Chat bubble fixes (canvas/src/components/tabs/ChatTab.tsx):
- User bubble: bg-accent-strong/30 + text-blue-100 → bg-accent + text-white
(translucent dark-blue overlay on warm-paper surface read as pale lavender
with near-invisible light-blue text — a real WCAG AA failure on the
highest-traffic surface in canvas).
- System/error bubble: bg-red-900/30 + text-red-200 → bg-bad/10 + text-bad,
using semantic tokens so dark-mode adapts automatically.
- Agent bubble: drop /80 + /30 opacity modifiers; solid bg-surface-card +
text-ink + border-line gives consistent contrast in both themes.
- prose-invert was unconditional, so markdown text on agent/system bubbles
rendered as light text on a light surface in light mode. Make it apply
only on the user bubble (the only dark surface in this component).
- Timestamp: text-ink-soft is too pale on warm-paper; use text-ink-mid for
agent/system, white/70 for user (visible on the now-solid blue bg).
Sub-tab bar fixes (canvas/src/components/SidePanel.tsx):
- Right-edge fade was hardcoded `from-zinc-950` — that paints a dark vertical
strip on the right edge of the tab bar in light mode. Switch to
`from-surface` so the gradient blends into whichever theme is active.
- Inactive tab text: text-ink-soft (~3.5:1 on warm-paper) → text-ink-mid
(~7:1). Active tab background: drop the /40 opacity so the selection is
unambiguous on either surface.
No semantic-token additions; all changes use existing warm-paper tokens
that already work in both themes.
Three loading-state divs were missing the role/aria pattern that
TemplatePalette.tsx and EmptyState.tsx already follow. Screen readers
get no signal that the page is waiting:
- canvas/src/app/page.tsx — full-screen "Loading canvas..." while
the websocket hydrates. First paint of the entire app.
- canvas/src/components/settings/TokensTab.tsx — "Loading tokens..."
- canvas/src/components/settings/OrgTokensTab.tsx — "Loading keys..."
Add role="status" + aria-live="polite" to the wrapping div so
assistive tech announces the wait and the eventual transition.
Visual rendering unchanged.
The tier system in CreateWorkspaceDialog and design-tokens has been
T1 Sandboxed / T2 Standard / T3 Privileged / T4 Full Access, but two
chrome surfaces still showed the older 3-tier mapping with T3 as
"Full Access":
- Legend (bottom-left chrome on every canvas page) listed only T1/T2/T3
and called T3 "Full Access". On a SaaS tenant the actual workspace
badges render T4 (in amber/warm) — there was no T4 entry in the
legend at all, so the user sees an undocumented orange badge.
- ConfigTab tier dropdown (per-workspace settings → Sandboxing) had no
T4 option at all and called T3 "Full Access". So an existing T4
workspace would show "T3 — Full Access" as the selected option,
silently downgrading the displayed tier on the settings panel.
- tenant.ts isSaaSTenant() doc comment claimed SaaS workspaces are
"inherently T3 Full Access" — wrong on both the number and the lock
rationale (SaaS hides T1/T2/T3, not just T1/T2).
Fix:
- Legend now imports TIER_CONFIG and renders all four tiers
(Sandboxed/Standard/Privileged/Full Access) using the same color
swatches as the badges on workspace cards. Eliminates the previous
drift where Legend's hardcoded sky/violet/warm chips didn't match
the gray/sky/violet/amber actually rendered on nodes.
- ConfigTab adds the missing T4 — Full Access option and renames T3
to Privileged.
- tenant.ts comment updated to match the picker's actual hide list.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
PR #2555 (Tailwind v4 + warm-paper) migrated all canvas chrome (toolbar,
side panel, modal layer) to semantic tokens, but missed the React Flow
viewport's `colorMode="dark"` literal — and two paired hardcoded dark
literals on the Background dot color and MiniMap mask. Net result on
prod: the user picked light mode, the toolbar flipped warm-paper, but
the canvas backplate, edges, dots, controls, and minimap stayed black —
visibly half-themed.
Three coordinated fixes inside the canvas viewport:
- ReactFlow `colorMode={resolvedTheme}` so the library's own dark/light
styles flip with the user's choice.
- Background dot color picks the line-soft tone in light mode (zinc-800
was invisible-on-cream).
- MiniMap maskColor warm-tints the off-viewport dim so the unselected
region doesn't render as a hard black bar over warm-paper.
Verification:
- `npx tsc --noEmit` clean
- `npx vitest run` 188/188 pass
- (will browser-verify post-redeploy on hongming.moleculesai.app)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Independent code review of #2555 caught two contrast regressions left
by the bulk perl pass:
1. text-white → text-ink mass-substitution silently broke destructive
and primary buttons. text-ink resolves to #15181c (warm-paper
near-black) in light mode — dark text on bg-red-600 / bg-amber-600
/ bg-emerald-600 / bg-blue-600 / bg-accent / bg-accent-strong /
bg-good / bg-bad fails WCAG contrast and looks broken. Per-line
pass flips text-ink → text-white only when a saturated bg utility
is present; tinted-state pills (bg-red-950/50 etc.) keep their
intentionally-retained text-* literals.
2. Original mapping table was missing bg-zinc-600 (most-used
hover-state literal for cancel buttons — caused them to JUMP from
warm cream resting state to dark zinc on hover in light mode) and
text-zinc-700/800/900 (separator dots and decorative dim text
invisible on warm-paper light bg). Extended mapping fills these
gaps with bg-surface-card / text-ink-soft.
Also: drop stale tailwind.config.ts reference from components.json
(file deleted by the v3→v4 migration); switch baseColor zinc →
neutral and enable cssVariables since v4 uses CSS-driven tokens.
Future shadcn-cli invocations would have failed or written malformed
components without this.
27 sites in 27 files affected by #1, ~20 sites in 20 files by #2.
1214/1214 unit tests still pass; build still clean.
Findings courtesy of multi-model review per code-review-and-quality
skill — different blind spots catch different bugs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
PR #2545 self-review findings.
(1) originalModel was set from wsMetadataModel alone. On a hermes/pre-#240
workspace where MODEL_PROVIDER was never written but YAML has
runtime_config.model: "something", originalModel="" while the form
rendered "something" — handleSave's diff fired /model PUT on every
unrelated save (tier change → workspace auto-restart). Snapshot from
the actual rendered model in BOTH loadConfig branches so the diff
stays scoped to user-initiated changes.
(2) The store-flush test asserted the call happened but didn't pin
success-gating. A future refactor wrapping the PATCH in try/catch and
unconditionally calling updateNodeData would have shipped green and
left the badge lying about server-rejected writes. New test pins the
PATCH-rejects-no-flush invariant.
(3) Hermes-edge regression test for (1).
All 1214 canvas tests pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three drift bugs in ConfigTab + ProviderModelSelector. Same root pattern:
the form's display, the diff baseline, and the canvas store all read or
write from different copies of the same data, so what the user sees and
what the runtime actually uses can diverge silently.
(1) currentModelId read runtime_config.model first; loadConfig overrode
only top-level config.model. With template YAML `runtime_config.model:
sonnet` and live MODEL_PROVIDER=`MiniMax-M2`, the form rendered
"Claude Code subscription / Claude Sonnet (OAuth)" while the container
env (and chat) used MiniMax-M2. Fix: loadConfig propagates
wsMetadataModel into BOTH places.
(2) handleSave's nextModel-vs-oldModel diff compared the form value to
the YAML default. After (1) mirrors wsMetadataModel into the form's
runtime_config.model for display, that diff was always non-zero on
no-op saves and would fire /model PUT — which auto-restarts. New
originalModel state tracks the loaded MODEL_PROVIDER and is the diff
baseline.
(3) handleSave PATCHed the workspace row but never pushed the same
fields into useCanvasStore.updateNodeData. User picked T3, hit Save &
Restart, DB updated to tier=3, header pill kept showing T2 until full
hydrate. Fix: mirror dbPatch into the store.
Bonus: ProviderModelSelector.handleProviderChange used to auto-default
the model to next.models[0] (alphabetically first) when switching
providers. User picked the MiniMax provider intending MiniMax-M2.7;
the form silently set MiniMax-M2 (first in the bucket) and the
workspace deployed with the wrong model. Now empty-default for
multi-model providers, force explicit pick — Save/Deploy already gate
on model.trim() === "".
Three new tests in ConfigTab.provider.test.tsx pin (1)/(2)/(3); two
existing ProviderModelSelector tests updated to reflect the no-silent-
default behaviour, with a new single-model-auto-pick test for the
0-vs-many boundary. 1212/1212 canvas tests pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The shared <ProviderModelSelector> component was authored on disk but
never landed — three deploy/configure surfaces still rendered the
legacy free-text "MODEL slug" input + provider-radio list. Tasks #239
and #243 closed at "component exists" rather than "user-visible
behavior changed", and the integration sat in a working-tree stash
that was never committed.
This PR is the missing integration:
- canvas/src/components/ProviderModelSelector.tsx (new, 509 lines):
single-source-of-truth Provider→Model cascade. Builds a catalog
from `template.models[].required_env` (groups by sorted+joined env
names so two MiniMax models with the same auth land in one
provider), exposes vendor detection helper + back-derivation. No
per-template hardcoding — fully driven by the upstream payload.
- canvas/src/components/MissingKeysModal.tsx: replaces the inline
`<input type="text">` + `<fieldset>` of provider radios with one
`<ProviderModelSelector>`. Same external contract
(`onKeysAdded(model)`), so callers in useTemplateDeploy don't move.
- canvas/src/components/tabs/ConfigTab.tsx: replaces ad-hoc Model
text input + Provider radio with the same selector, fixing the
display-vs-storage drift class that #190 first patched.
Tests
=====
- ProviderModelSelector.test.tsx (new, 269 lines): cascade behavior,
vendor auto-snap, back-derivation from saved config.
- MissingKeysModal.cascade.test.tsx: rewritten to assert dropdown
shape (was asserting the legacy text-input shape).
- ConfigTab.hermes.test.tsx + ConfigTab.provider.test.tsx: updated
for the new selector shape.
- 1208/1208 canvas tests pass locally.
User-visible fix: clicking any deploy/configure surface from the
sidebar now shows the cascade UX (Provider dropdown first, Model
dropdown filtered) instead of the legacy free-text MODEL slug.
Closes the integration gap behind #239 + #243. Builds on merged
runtime PRs #2538 (universal MODEL_PROVIDER) + #32 + #38 (per-vendor
audit).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The TemplatePalette deploy modal (MissingKeysModal → ProviderPickerModal)
let the model field and provider radio drift apart. When a hermes
template defaulted the model to "MiniMax-M2.7-highspeed" but the radio
defaulted to providers[0] (Anthropic), the env-var input below asked
for ANTHROPIC_API_KEY. A user pasting their MINIMAX_API_KEY there (or
just dismissing the dialog) ended up with a workspace whose
runtime_config.model=MiniMax + ANTHROPIC_API_KEY env — the hermes
adapter then crashed during boot before /registry/register, surfacing
as WORKSPACE_PROVISION_FAILED 12 minutes later.
Caught 2026-05-02 on hongming/Hermes Agent (workspace 95ed3ff2-…
ended with: "container started but never called /registry/register").
Sibling of the ConfigTab cascade fix in PR #2516 (task #236) — same
pattern, different surface. Plumbs the template's full ModelSpec[]
(with required_env per model) into the picker. When the typed model
matches a registry entry, snap the radio so the env-var fields
underneath match what the model actually needs.
Free-text models (typed slug not in the registry) and models with no
required_env (local/self-hosted endpoints) leave the radio alone — the
user can still pick a provider manually. Backwards-compat: callers
that don't pass `models` get the pre-cascade behavior, pinned by a
regression test.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Follow-up to PR #2509/#2510. The defensive v1-detection branches in
extract_attached_files (Python) and extractFilesFromTask (TypeScript)
were merged with comments claiming they fix a "v0→v1 silent-drop"
bug that surfaced as the 2026-05-01 hongming "no text content"
incident. Live test disproved that hypothesis: a2a-sdk's JSON-RPC
layer validates inbound requests against the v0 Pydantic union, so
v1 shapes are rejected at the request boundary — the v1 detection
branch is unreachable on the JSON-RPC ingress path. The actual root
cause of the hongming incident was the missing /workspace chown
fixed by CP PR #381 + test #382.
Update the comments to honestly describe these branches as
defensive future-proofing (kept against an eventual SDK schema
migration or in-process callers that construct Parts directly from
protobuf), not as fixes for an observed bug. Also trims
ChatTab.tsx's outbound-shape comment block from ~21 lines to a
3-line pointer to the SDK union.
Comment-only change. No behavior change. 86 workspace tests + 91
canvas tests still pass.
The previous PR (#2509) flipped canvas outbound file parts to the v1
flat shape `{url, filename, mediaType}` based on a hypothesis that
a2a-sdk's JSON-RPC parser silently dropped v0 `{kind:"file", file:{...}}`
shapes. Live test shows the opposite: a2a-sdk's JSON-RPC layer
validates against the v0 Pydantic discriminated union (TextPart |
FilePart | DataPart), so v1 flat shape is rejected with:
Invalid Request:
params.message.parts.0.TextPart.text — Field required
params.message.parts.0.FilePart.file — Field required
params.message.parts.0.DataPart.data — Field required
The actual root cause of the user-visible "Error: message contained
no text content" was the missing `/workspace` chown (CP PR #381 +
test pin #382), not a wire-shape mismatch. Verified end-to-end by
sending a v0 image-only message after PR #381 + workspace re-provision
— agent receives the file, reads its bytes, and replies normally.
Reverting only the canvas outbound shape. Defensive v1-tolerance
stays in:
- workspace/executor_helpers.py — extract_attached_files still
accepts v1 protobuf parts in case a future client emits them or
a future SDK release flips internal representation. Harmless on
the v0 hot path.
- canvas/message-parser.ts — extractFilesFromTask still tolerates
v1 shape on incoming agent responses. Some agents may emit v1
when their internal serializer round-trips through protobuf.
Tests stay green (91 canvas, 86 workspace).
Image-only chats surface "Error: message contained no text content"
because canvas posts v0 `{kind:"file", file:{uri,name,mimeType}}` shapes
that the workspace runtime's a2a-sdk v1 protobuf parser silently drops:
v1 `Part` has fields `[text, raw, url, data, metadata, filename,
media_type]` and `ignore_unknown_fields=True` discards `kind`+`file`,
producing a fully-empty Part. With no text and no extracted file
attachments, the executor's "no text content" guard fires.
Three coordinated changes close the gap:
1. canvas/ChatTab.tsx — outbound file parts now carry the v1 flat
shape `{url, filename, mediaType}` so the v1 protobuf parser
populates Part fields instead of dropping them.
2. workspace/executor_helpers.py — extract_attached_files learns the
v1 detection branch (non-empty `part.url` + `filename` +
`media_type`) alongside the existing v0 RootModel and flat-file
shapes. Defends every runtime that mounts the OSS wheel against
the same drop, including any pre-fix client still on the wire.
3. canvas/message-parser.ts — extractFilesFromTask tolerates the v1
shape on incoming agent responses too, so file chips render in
chat history regardless of which Part shape the runtime emits.
Test pins:
- workspace/tests/test_executor_helpers.py:
+ v1 protobuf shape extraction
+ empty-Part defense (v0→v1 silent-drop fall-through returns [])
- canvas message-parser test:
+ v1 protobuf flat parts
+ filename fallback to URL basename for v1
Self-review of #2460 found two issues:
1. Critical: Override button in ProviderPickerModal called
/settings/secrets when no workspaceId, overwriting the GLOBAL
secret used by every workspace. The only consumers of this
modal today (TemplatePalette, EmptyState via useTemplateDeploy)
never pass workspaceId, so Override was always destructive.
Removed entirely — the picker still solves the user-reported
bug (always-ask + reuse saved keys); per-workspace key override
can be a separate PR that plumbs secrets through POST /workspaces.
2. Optional: /settings/secrets was being fetched twice — once
inside checkDeploySecrets (silently) and again in the hook to
populate configuredKeys. Surfaced configuredKeys on
PreflightResult so the hook re-uses the existing fetch.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Clicking a hermes template tile silently deployed when global env
covered the API key, producing "No LLM provider configured" 500
because the workspace booted with no explicit model slug — the
adapter fell back to its compiled-in default which 401s on the
user's actual provider key.
Fix: in useTemplateDeploy, open the picker whenever the template
declares ≥2 provider options, even when preflight.ok=true. The
modal renders pre-saved keys as Saved (with an Override link) and
adds a model input pre-filled from the template's default. Single-
provider templates (claude-code, langgraph) still skip the picker
since there's nothing to choose.
POST /workspaces now includes the picker's model slug so hermes-
style routing reads the prefix at install time.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Mirrors the data-driven pattern PR #2454 set in ConfigTab: read
runtime_config.providers from /templates and filter the modal's
provider <select> to that subset. Same source of truth, three fewer
hardcoded copies of the provider list.
Behavior:
- Template declares providers → dropdown shows only those.
- Template ships no providers field → fall back to full HERMES_PROVIDERS
catalog (back-compat for older templates / self-hosted setups).
- Declared list has no overlap with our static metadata → fall back to
full catalog so the form can't lock the operator out.
- hermesProvider snaps back to the first available pick when its
current value falls out of the filtered list.
Tests: 3 new pinning the filter, no-providers-field fallback, and
the unknown-providers fallback. All 27 CreateWorkspaceDialog tests
pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Option B PR-5. Canvas Config tab now exposes a Provider override input
that's adapter-driven from each runtime's template — no hardcoded
provider list in the canvas. PUT /workspaces/:id/provider on Save
when dirty; auto-restart suppression to avoid double-restart with
the model handler's own restart.
The dropdown's suggestion list comes from /templates →
runtime_config.providers (the field added in
molecule-ai-workspace-template-hermes PR #31). For templates that
haven't migrated to the explicit providers list yet, suggestions
derive from model[].id slug prefixes — still adapter-driven, just
inferred. This keeps existing templates working while platform team
migrates them one at a time.
workspace-server changes:
- Add Providers []string field to templateSummary JSON
- Parse runtime_config.providers in /templates handler
- 2 new tests pin the surfacing + omitempty behavior
canvas changes:
- Remove hardcoded PROVIDER_SUGGESTIONS constant
- Add provider/originalProvider state + PUT-on-save logic
- Add deriveProvidersFromModels() fallback helper
- Wire RuntimeOption.providers from /templates response
- 8 new tests pin the behavior end-to-end
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The Model dropdown's onChange writes to config.runtime_config.model
whenever a runtime is set (hermes, claude-code, etc.), and only
falls back to top-level config.model when no runtime is selected.
But handleSave used to diff the new value against top-level
nextSource.model only — so for any runtime-bearing workspace, the
PUT /workspaces/:id/model never fired and MODEL_PROVIDER never
landed in workspace_secrets.
Symptom (2026-04-30, hongmingwang Hermes Agent
32993ee7-840e-4c02-8ca8-cb9d75d112a5):
- User picks minimax/MiniMax-M2.7-highspeed from the dropdown
- Hits Save & Restart
- Save reports success; restart fires
- The new EC2 boots with HERMES_DEFAULT_MODEL empty
- install.sh defaults to nousresearch/hermes-4-70b
- hermes-agent errors "No LLM provider configured" on every chat
turn because no NOUS_API_KEY / OPENROUTER_API_KEY is set
- Reload Config tab → model field reverts to whatever
GET /workspaces/:id/model returns (i.e. empty / template default)
handleSave now reads the effective model from runtime_config.model
first and falls back to top-level model for legacy no-runtime
workspaces. Same change for the old-value diff so a no-op Save
still skips the PUT.
Tests pin both branches: PUTs /model when the dropdown changed
runtime_config.model on a hermes workspace; does NOT PUT when
the value is unchanged from what GET /model returned.
Critical:
- ExternalConnectModal.tsx: filledUniversalMcp substitution searched
for WORKSPACE_AUTH_TOKEN but the snippet's placeholder is now
MOLECULE_WORKSPACE_TOKEN (changed in the previous polish commit
876c0bfc). Operators copy-pasting the MCP tab would have gotten a
literal "<paste from create response>" instead of the token. Fix
the substitution to match the new placeholder name.
Important:
- mcp_cli._platform_register: 401/403 from initial register now hard-
exits with code 3 + an actionable stderr message pointing the
operator at the canvas Tokens tab. Pre-fix: warning log + continue,
which made a bad-token startup silently fail (heartbeat 401's
forever, every tool call also 401's, no clear surfacing in the
operator's MCP client). 500/503 still log + continue (transient
platform blips shouldn't abort the MCP loop).
- a2a_mcp_server.cli_main docstring: removed stale claim that this is
the wheel's console-script entry-point target. The actual target is
mcp_cli.main since 2026-04-30. Wheel-smoke pins both names so the
functionality was correct, but the doc was lying.
Test coverage: 3 new mcp_cli tests:
- register 401 exits code=3 + stderr mentions canvas Tokens tab
- register 403 (C18 hijack rejection) takes same path
- register 500/503 does NOT exit — only auth errors hard-fail
Findings deferred to follow-up (acceptable per review rubric):
- Code dedup across mcp_cli / heartbeat.py / molecule_agent SDK
- Pooled httpx.Client for connection reuse
- Heartbeat exponential backoff
- Token-resolution ordering parity (env-first vs file-first)
between mcp_cli.main and platform_auth.get_token
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The canvas tab snippet for the Universal MCP path was written before
this PR added the built-in register + heartbeat thread. Earlier wording
described it as "outbound-only — pair with the Claude Code or Python SDK
tab for heartbeat + inbound messages" — that's stale. molecule-mcp now
handles register + heartbeat itself; the only thing it doesn't yet do is
inbound A2A delivery.
Updated:
- externalUniversalMcpTemplate header comment + body — describes
standalone behavior, points operators at SDK/channel only when they
need INBOUND (not heartbeat).
- Drops the now-redundant curl-register step from the snippet — the
binary registers itself on startup.
- Canvas modal label likewise updated.
No runtime / behavior change; pure docs polish so a copy-pasting
operator's mental model matches what the binary actually does.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The "Connect your external agent" dialog already covered Claude Code,
Python SDK, curl, and raw fields. This adds a Universal MCP tab that
documents the new \`molecule-mcp\` console script — the runtime-
agnostic baseline shipped by PR #2413's workspace-runtime changes.
Surface area:
- New \`externalUniversalMcpTemplate\` constant in workspace-server.
Three-step snippet: pip install runtime → one-shot register via curl
→ wire molecule-mcp into agent's MCP config (Claude Code example,
notes that hermes/codex/etc. take the same env-var contract).
- Workspace create response now includes \`universal_mcp_snippet\`
alongside the existing curl/python/channel snippets.
- Canvas modal renders the tab when \`universal_mcp_snippet\` is
present; backward-compatible with older platform builds (tab hides
when empty).
Origin/WAF coverage (the user explicitly asked for this):
- The runtime wheel handles Origin automatically (this PR's earlier
commit on platform_auth.auth_headers).
- The curl tab now sets \`Origin: {{PLATFORM_URL}}\` preemptively
with an explanatory comment; \`/registry/register\` is currently
WAF-allowed without it but adding now keeps the snippet working
if WAF rules expand. The comment also explains why
\`/workspaces/*\` paths return empty 404 without Origin — the
exact failure mode I hit while smoke-testing this PR live.
- The MCP snippet's footer notes that the wheel auto-handles
Origin so operators don't think about it.
End-to-end verification (against live tenant
hongmingwang.moleculesai.app, freshly registered workspace):
- get_workspace_info → full JSON
- list_peers → "Claude Code Agent (ID: 97ac32e9..., status: online)"
- recall_memory → "No memories found."
all returned by the molecule-mcp binary speaking MCP stdio to
this Claude Code session.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When the platform's create-external-workspace response includes
`claude_code_channel_snippet` (added in this same PR's first commit),
the modal surfaces it as the **first** tab — defaulting to it for new
external workspaces because polling-based + no-tunnel is the lowest-
friction path. Falls back to Python tab when the field is absent
(older platform builds).
Type addition is optional (`claude_code_channel_snippet?: string`)
so the canvas keeps building against pre-#2304 platform responses
during the soak window.
Auth-token stamping mirrors existing python/curl behavior — the
.env's `MOLECULE_WORKSPACE_TOKENS=<paste auth_token from create
response>` placeholder gets filled in client-side so the copy-paste
block is truly ready to run.
Also adds the missing 'use client' directive — the file uses useState
+ useCallback but didn't have the Next.js client-component marker.
Pre-commit caught it; existing absence was a latent bug that would
surface as an SSR hook error if any path rendered this component
during server rendering.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Both AgentCommsPanel and ChatTab's activity-feed opened raw
`new WebSocket(WS_URL)` instances per mount, with no onclose handler
and no reconnect logic. When the underlying connection dropped — idle
timeout, browser background-tab throttle, network jitter — the per-
panel sockets stayed dead until the panel re-mounted (refresh or
sub-tab unmount/remount). Live agent-comms bubbles and live activity
feed lines silently went missing in the gap, manifesting as "the
delegation didn't show up until I refreshed."
The global ReconnectingSocket in store/socket.ts already owns
reconnect, exponential backoff, health-check, and HTTP fallback poll.
Routing component subscribers through it gives every consumer those
guarantees for free, with one TCP connection per tab instead of N.
Three new pieces:
- store/socket-events.ts: tiny pub/sub bus. emitSocketEvent fan-outs
every decoded WSMessage to the listener Set; subscribeSocketEvents
returns an unsubscribe. A throwing listener is logged and isolated
so it can't break siblings.
- store/socket.ts: ws.onmessage now calls emitSocketEvent(msg) right
after applyEvent(msg), so the store's derived state and component
subscribers stay in lockstep on every event arrival.
- hooks/useSocketEvent.ts: React hook that registers exactly once
per mount, capturing the latest handler in a ref so the closure
sees current state/props without re-subscribing on every render.
Refactored sites:
- AgentCommsPanel: replaced its WebSocket-in-useEffect block with
useSocketEvent. Same parsing logic; the panel no longer opens its
own connection.
- ChatTab activity feed: split the previous useEffect in two — one
seeds the activity log when `sending` flips, the other subscribes
unconditionally and gates work on `sending` inside the handler.
Hooks can't be conditional, so the gate has to live in the body
rather than around the effect.
The ws-close graceful-close helper is no longer needed in either
site; the global socket owns its own teardown.
Tests: 6 new tests for the bus contract (single delivery, fan-out
order, unsubscribe, throwing-listener isolation, no-subscriber emit,
duplicate-subscribe Set semantics). All 27 existing socket tests
still pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Third-pass review caught a fourth WS path I missed. The original fix +
the stale-callback follow-up patched 3 sites that release the in-flight
guards (pendingAgentMsgs effect, HTTP .then() success, HTTP .catch()
success), but the ACTIVITY_LOGGED handler at lines 410-419 also clears
`sending` + `sendingFromAPIRef` when the platform logs the workspace's
a2a_receive ok/error. It only cleared 2 of the 3 refs — same exact
bug class as the original. If THIS path wins the race (a2a_receive
activity logged before pendingAgentMsgs delivers the reply text),
sendInFlightRef stays stuck true and the next sendMessage() silently
no-ops at line 464.
Fix: route both branches (ok and error) through releaseSendGuards()
so all four sites are now uniform.
Updated the helper's docstring to explicitly list all four sites and
warn that any future "I saw the reply" path that only clears the
natural pair (sending + sendingFromAPIRef) will silently re-introduce
the freeze. The disabled-button logic can't see sendInFlightRef so
the visible state diverges from the synchronous re-entry guard
otherwise.
This is exactly the drift `releaseSendGuards()` was supposed to
prevent — the helper landed in the prior commit but the activity-log
site wasn't migrated to use it. Fixing now closes the gap.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Self-review on PR #2185 surfaced a latent race the original fix exposed:
the WS-clears-guards path now releases sendInFlightRef immediately, which
means a user can fire msg #2 between WS-arrival and HTTP-arrival for
msg #1. Without coordination, msg #1's late .then() sees
sendingFromAPIRef=true (set by msg #2's send), enters the main body,
and runs setSending(false) + appendMessageDeduped against msg #1's
response body — clobbering msg #2's in-flight UI state.
This race is realistic for claude-code SDK: the comment at line 294-298
already calls WS the "authoritative reply arrived" signal, and the user
typically reads-then-types before the trailing HTTP completes. Without
the original Send-button freeze "protecting" the race, it surfaces.
Two changes:
1. Token-keyed callbacks. sendTokenRef bumps on every sendMessage
entry; .then()/.catch() capture the token in closure and bail
without touching any flags if a newer send has superseded them.
The newer send owns the in-flight guards.
2. releaseSendGuards() helper. The three-clear-guards trio
(setSending, sendingFromAPIRef, sendInFlightRef) now lives in one
useCallback so the WS handler, .then() success, and .catch()
success can't drift apart. A future contributor dropping one of
the three would silently re-introduce either the post-WS Send
freeze or the stale-callback clobber.
Skipped a unit test for this regression — ChatTab has no __tests__
file and a mount test would need WS + zustand + api mocks. The fix
is 4 logical lines (token capture + 2 guard checks) and the manual
test covers it. Follow-up to add a focused mount test when ChatTab
gets its first __tests__ file.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Send button + Enter both silently no-op'd after the first agent reply
on runtimes that deliver via WebSocket (claude-code SDK does this per
the comment at ChatTab.tsx:294-298). The visible disabled-state checks
(sending, uploading, agentReachable) were all clean — the freeze came
from a third synchronous reentry guard the button can't see:
if (sendInFlightRef.current) return; // ChatTab.tsx:438
The ref was set true at the start of sendMessage() and only cleared in
.then() / .catch() of the HTTP fall-through and the upload-failure
branch. The WS-push handler in the pendingAgentMsgs effect cleared
`sending` and `sendingFromAPIRef` but left `sendInFlightRef` stuck
true. The HTTP .then() then early-returned at the dedup check (line
513) without touching the ref — only the .catch() early-return path
did. Net result: refresh fixed it because the ref reset on remount.
Two-line fix:
- WS handler: also clear sendInFlightRef when the push delivers the
reply (primary fix; no race window where the ref is stuck while
the user can already type)
- .then() early-return: mirror .catch()'s cleanup as defense in
depth, so neither delivery order leaks the ref
While here: A2AEdge.test.tsx fixture was typed `as never` to dodge
EdgeProps' discriminated-union complaint, which broke spreading at
the call sites with TS2698 ("Spread types may only be created from
object types"). Replaced with `as unknown as ComponentProps<typeof
A2AEdge>` — preserves the original "skip restating every optional
field" intent and keeps a spreadable type.
All 10 A2AEdge tests pass; tsc --noEmit is clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
[Molecule-Platform-Evolvement-Manager]
Closes the fourth and final item from #2071 — but at a slightly
different layer than the issue listed: tests `dragUtils.ts` (the
74-LOC pure-ish geometry helpers) instead of the full 296-LOC
`useDragHandlers` hook. Rationale below.
15 cases across 2 buckets:
**shouldDetach (8):**
- child fully inside parent → false
- child drifted slightly past edge but under DETACH_FRACTION → false
- child past 20% threshold on X → true (un-nest)
- child past 20% threshold on Y → true (un-nest)
- missing child node → true (conservative fallback per source comment)
- missing parent node → true (same)
- measured size absent → falls back to React Flow's 220x120 defaults
(mirrors initial-mount race where measurement hasn't run yet)
- DETACH_FRACTION constant pinned at 0.2 (Miro/tldraw convention)
**clampChildIntoParent (7):**
- child already inside bounds → no-op (no setState — proven by
reference equality on mockState.nodes)
- drifted past top-left → clamps to (0, 0)
- drifted past bottom-right → clamps to (parentW - childW, parentH - childH)
- per-axis independence: X past edge + Y inside → only X clamps
- child not in store → early return, no setState
- child internalNode missing → early return, no setState
- multi-node store: clamping one node MUST NOT touch siblings
## Why dragUtils, not the full useDragHandlers hook
The hook (296 LOC) orchestrates React Flow drag events + Zustand
mutations. Testing it would need heavyweight `useReactFlow` +
internal-node + `setDragOverNode` / `nestNode` / `batchNest` /
`isDescendant` mocks just to drive event handlers — and the
*decisions* the hook makes all delegate to these two helpers:
- `shouldDetach` decides "is this a real un-nest?"
- `clampChildIntoParent` snaps the child back when the user drifted
slightly past the edge without holding Alt/Cmd
Pinning these locks the hot path the user feels. The hook's
remaining surface (modifier-key snapshotting, drop-target
broadcasting, commit-on-release grow pass) is plumbing — worth
testing as a follow-up if it ever regresses, but lower
correctness leverage per LOC of test setup.
## #2071 status after this PR
- [x] useTemplateDeploy (#2121)
- [x] A2AEdge (#2143)
- [x] OrgCancelButton (#2145)
- [x] dragUtils geometry helpers (this PR)
- [ ] Full useDragHandlers hook orchestration — explicit deferral
with rationale above
## Test plan
- [x] All 15 cases pass locally (`vitest run dragUtils.test.ts` — 131ms)
- [x] No changes to the SUT — pure additive coverage
- [ ] CI green
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
[Molecule-Platform-Evolvement-Manager]
Closes the third item from #2071 (Canvas test gaps follow-up). Builds
on the A2AEdge tests in PR #2143.
10 cases across 4 buckets:
**Render (2):**
- Default pill with `Cancel (N)` text + correct ARIA label
- Confirm dialog NOT visible until pill click
**Pill click (3):**
- Click flips to confirming view + stops propagation (so React Flow
doesn't interpret the click as a node selection)
- Confirm copy pluralizes correctly: count=1 → "Delete 1 workspace?",
count>1 → "Delete N workspaces?". Negative assertion guards against
the wrong-form regressing in either direction.
**No / cancel-confirm (1):**
- Click No → returns to pill, no API call, no store mutation
**Yes / cascade-delete (4):**
- Happy path: beginDelete locks the WHOLE subtree (root + children,
NOT unrelated workspace) → api.del("/workspaces/<id>?confirm=true")
→ optimistic store filter strips subtree, keeps unrelated → success
toast → endDelete in finally
- WS-event race: WS_REMOVED handler clears the root mid-flight. The
bail-out branch (`!postDeleteState.nodes.some(n => n.id === rootId)`)
must NOT then run a second optimistic filter. Pre-fix the post-await
subtree walk would miss any orphaned descendants whose parentId got
reparented upward by handleCanvasEvent — pinned now.
- Error path: api.del rejects → endDelete UNDOes the lock + error
toast surfaces the message → subtree STAYS in the store so the user
can retry / interact with the still-deploying nodes
- Non-Error rejection (e.g. string thrown directly): toast surfaces
the canned "Cancel failed" fallback instead of attempting `.message`
## Mocking
- `@/lib/api`, `@/components/Toaster`: simple spy mocks
- `@/store/canvas`: object that satisfies BOTH the selector pattern
(`useCanvasStore(s => s.x)`) AND `getState()` / `setState()` since
the cascade-delete handler walks the subtree via `getState()` and
mutates via `setState()` for the optimistic removal. `vi.hoisted`
preserves referential identity so the mock fns wired into the
state object are observed by every consumer.
## Test plan
- [x] All 10 cases pass locally (`vitest run OrgCancelButton.test.tsx` — ~990ms)
- [x] No changes to the SUT — pure additive coverage
- [ ] CI green
## #2071 progress after this PR
- [x] useTemplateDeploy (PR #2121)
- [x] A2AEdge (PR #2143)
- [x] OrgCancelButton (this PR)
- [ ] useDragHandlers — separate PR
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
[Molecule-Platform-Evolvement-Manager]
Closes the second item from #2071 (Canvas test gaps follow-up):
adds behavioural coverage for the custom React Flow edge that renders
delegation counts between workspaces and routes a click into the
source workspace's Activity feed.
10 cases across 2 buckets:
**Render (6):**
- Empty label → BaseEdge only, NO portaled HTML pill (the most
common state for cold edges; pill must not render-through-empty)
- Non-empty label → pill renders with the exact label text
- isHot=true → violet accent classes; blue accent NOT present
- isHot=false → blue accent classes
- ARIA pluralization: count=1 → "1 delegation from …" (singular)
- ARIA pluralization: count=7 → "7 delegations from …" (plural)
**Click behaviour (4):**
- Click → selectNode(source)
- FRESH selection (selectedNodeId != source) → also setPanelTab("activity")
- RE-click of already-selected source → setPanelTab MUST NOT fire
(this is the regression-locked guarantee — preserves the user's
current tab when they intentionally moved to Chat / Memory while
inspecting the same peer)
- stopPropagation: parent onClick must NOT see the event (otherwise
the canvas Pane's clear-selection handler would fire and undo the
edge's own selectNode call)
## Mocking strategy
- `@xyflow/react`: BaseEdge → <g data-testid>, EdgeLabelRenderer →
inline pass-through (no portal), getBezierPath → fixed [path, x, y].
Lets the test render the component without a ReactFlow provider.
- `@/store/canvas`: vi.hoisted-shared mock state with selectNode +
setPanelTab spies and a mutable selectedNodeId. The store's
getState() returns the same object so the click handler's
`useCanvasStore.getState().selectedNodeId` lookup works.
Pattern matches the existing `A2ATopologyOverlay.test.tsx` setup
in the same module.
## Test plan
- [x] All 10 cases pass locally (`vitest run A2AEdge.test.tsx` — ~1.3s)
- [x] No changes to the SUT — pure additive coverage
- [ ] CI green
## Remaining #2071 items
- OrgCancelButton tests
- useDragHandlers tests
Each is a separate PR.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Reviewer bot flagged: ChatTab.tsx imported extractResponseText but
no longer used it after the loop body moved to historyHydration.ts
(the helper imports it directly). Drop from the named import to
unblock merge. extractFilesFromTask remains used at line 515 for the
WS A2A_RESPONSE handler's reply-files extraction.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Reviewer follow-up to PR #2134 (Optional finding). The history loader
walked text on the user branch but never extracted file parts — so a
chat reload after a session where the user dragged in a file rendered
the text bubble but lost the download chip. Symmetric to the agent
branch which already handles this via extractFilesFromTask.
Wire shape from ChatTab's outbound POST:
request_body = {params: {message: {parts: [
{kind: "text", text: "..."},
{kind: "file", file: {uri, name, mimeType?, size?}}
]}}}
extractFilesFromTask walks `task.parts`, so we feed it `params.message`
(the inner object that has the parts array). Three new tests:
- hydrates file attachments from request_body
- emits an attachments-only bubble when text is empty (drag-drop
without caption — pre-fix the empty userText short-circuited and
the row was dropped entirely)
- internal-self predicate suppresses the row even with attachments
(defence-in-depth for future internal triggers)
Stacked on #2134; this branch's parent commit is its tip.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
User pushed back: the timestamp bug should have been caught by E2E.
Right — my earlier coverage tested the server contract (notify endpoint,
WS broadcast filter) but never the chat-history HYDRATION path. Without
a unit test that froze the wall clock and asserted timestamps came from
created_at, a future refactor could re-introduce the same bug.
This commit:
1. Extracts the per-row → ChatMessage[] mapping out of the closure
inside loadMessagesFromDB into chat/historyHydration.ts. Pure
function, no React dependency, easy to test.
2. Adds 12 vitest cases in __tests__/historyHydration.test.ts covering:
- Timestamp regression (3 tests, with system time frozen to 2030 so
a regression starts producing "2030-…" timestamps and the assertion
fails unmistakably). The third test mirrors the user's screenshot:
two rows with distinct created_at must produce distinct timestamps.
- User-message extraction (text, internal-self filter, null body)
- Agent-message extraction (text, error→system role, file attachments,
null body, body with neither text nor files)
- End-to-end: a single row with both request and response emits
two messages with the same timestamp (the canonical canvas-source
row pattern)
3. The new file-attachment test caught a SECOND latent bug — the helper
was passing `response_body.result ?? response_body` to extractFiles
FromTask, which passes the STRING "<text>" for the notify-with-
attachments shape `{result: "<text>", parts: [...]}` and silently
returns []. So a chat reload after an agent attached a file would
lose the chips. Fixed by only unwrapping `result` when it's an
object (the task-shape) and falling through to response_body
otherwise (the notify shape).
ChatTab now imports the helper and the loop body becomes one line:
`messages.push(...activityRowToMessages(a, isInternalSelfMessage))`.
Verification:
- 12/12 historyHydration tests pass
- 1072/1072 full canvas vitest pass (was 1060 before, +12)
- tsc --noEmit clean
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
User flagged that all historical user bubbles render with the same
"now" clock after a chat reload — both messages in the screenshot
showed 9:01:58 PM despite being sent hours apart.
ChatTab.tsx:142 minted user messages with createMessage(...) which
calls new Date().toISOString() — fine for a freshly-typed message,
wrong for hydrated history. Every reload re-stamped all user bubbles
to the render moment, collapsing the visible chronology. The agent
path on line 157 already overrides with a.created_at; mirror that.
One-line fix (spread + override timestamp) plus a comment explaining
why the override is load-bearing so the next refactor doesn't drop it.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
User asked to "keep optimizing and comprehensive e2e testings to prove all
works as expected" for the communication path. Adds three layers of coverage
for PR #2130 (agent → user file attachments via send_message_to_user) since
that path has the most user-visible blast radius:
1. Shell E2E (tests/e2e/test_notify_attachments_e2e.sh) — pure platform test,
no workspace container needed. 14 assertions covering: notify text-only
round-trip, notify-with-attachments persists parts[].kind=file in the
shape extractFilesFromTask reads, per-element validation rejects empty
uri/name (regression for the missing gin `dive` bug), and a real
/chat/uploads → /notify URI round-trip when a container is up.
2. Canvas AGENT_MESSAGE handler tests (canvas-events.test.ts +5) — pin the
WebSocket-side filtering that drops malformed attachments, allows
attachments-only bubbles, ignores non-array payloads, and no-ops on
pure-empty events.
3. Persisted response_body shape test (message-parser.test.ts +1) — pins
the {result, parts} contract the chat history loader hydrates on
reload, so refreshing after an agent attachment restores both caption
and download chips.
Also wires the new shell E2E into e2e-api.yml so the contract regresses
in CI rather than only in manual runs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
User flagged two paper cuts in Agent Comms after the grouping PR:
"Delegating to f6f3a023-ab3c-4a69-b101-976028a4a7ec" reads as gibberish
because it's a UUID, and the chat is "one way" with only outbound bubbles
even though peers are clearly responding.
Both fixes are in toCommMessage's delegation branch:
1. Pull text from the actual payload, not the platform's audit-log summary.
- delegate row → request_body.task (the task text the agent sent).
Fallback when missing: "Delegating to <resolved-peer-name>" — never
the raw UUID.
- delegate_result row → response_body.response_preview / .text (the
peer's actual reply). Fallback paths render human-readable status
for queued / failed cases ("Queued — Peer Agent is busy on a prior
task...") instead of platform jargon.
2. delegate_result rows render flow="in" — even though source_id=us
(the platform writes the row on our side), the conversational
direction is peer → us. The chat now shows alternating bubbles
(out: "Build me 10 landing pages" → in: "Done — ZIP at /tmp/...")
instead of one-sided "→ To X" wall.
The WS push handler in this same file now populates request_body /
response_body from the DELEGATION_SENT / DELEGATION_COMPLETE event
payloads (task_preview, response_preview), so live-pushed bubbles use
the same text-extraction path as the GET-on-mount.
Tests:
- 4 new in toCommMessage's delegation branch:
- delegate row prefers request_body.task over summary
- delegate row falls back to name-resolved label when task missing
- delegate_result row is INBOUND (flow="in")
- delegate_result queued shows human-readable wait message including
the resolved peer name
- Replaces the previous "delegate row maps text from summary" tests
which encoded the (now-undesirable) platform-summary-as-text behavior.
- All 15 tests pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The chronological-only view was a noodle once Director + N peers
exchange more than a few rounds. New layout: a sub-tab bar at the
top of the panel, with "All" pinned leftmost and one tab per peer
(name + count). Selecting a peer filters the thread to that one
DD↔X conversation; "All" preserves the previous chronological view
as the default.
Tab ordering follows Slack/Linear DM-list convention: most-recent
activity descending, so active conversations rise to the top
without the user scrolling. Counts in parens match Slack's unread
hint pattern (no separate read/unread state — the count is total
in this conversation, computed from the same in-memory message
list the panel already maintains).
Pure-helper extraction: peer-summary derivation lives in
`buildPeerSummary(messages)` so the sort + count logic is unit-
testable without rendering the panel. 5 new tests cover: count
aggregation, most-recent-first ordering, lastTs as max-not-last,
empty input, name-stability when the same peerId carries different
names across messages.
Keyboard: ArrowLeft/Right cycle peer tabs (matches the existing
My Chat / Agent Comms tab pattern in ChatTab). Auto-prune: if the
selected peer has zero messages after a setMessages update (rare,
e.g. dedupe drops the last bubble), fall back to "All" so the
viewer doesn't see an empty thread.
Frontend-only — no platform / runtime / DB changes. The existing
`peerId` / `peerName` fields on CommMessage already carry every
piece of data the new UI needs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
[Molecule-Platform-Evolvement-Manager]
Addresses github-code-quality finding on PR #2064:
> Comparison between inconvertible types
> Variable 'info' cannot be of type null, but it is compared to
> an expression of type null.
By line 75, `info` has been narrowed to non-null via the
`if (!info) return null;` guard at line 56 — so `open={info !== null}`
always evaluates to `true`. Switch to JSX shorthand `open` for
clarity and to silence the static check.
Behaviorally identical; the modal still opens whenever the parent
renders this component (which only happens with non-null info).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Critical follow-up to PR #2126's review. Two real bugs:
1. **Runtime QUEUED never resolved.** Platform's drain stitch updates
the platform's delegate_result row when a queued delegation finally
completes, but never pushes back to the runtime. The LLM polling
check_delegation_status saw status="queued" forever — combined with
the new docstring guidance ("queued → wait, peer will reply"), the
model would wait indefinitely on a state that never resolves.
Strictly worse than pre-PR behavior where it would have at least
bypassed.
2. **Live updates dead code.** delegation.go writes activity rows by
direct INSERT INTO activity_logs, bypassing the LogActivity helper
that fires ACTIVITY_LOGGED. Adding "delegation" to the canvas's
ACTIVITY_LOGGED filter (PR #2126 first cut) was inert — initial
GET worked, live updates did not.
Fix:
(1) Runtime side, workspace/builtin_tools/delegation.py:
- New `_refresh_queued_from_platform(task_id)` async helper that
pulls /workspaces/<self>/delegations and finds the platform-side
delegate_result row for our task_id.
- check_delegation_status calls _refresh when local status is
QUEUED, so the LLM's poll itself drives state convergence.
- Best-effort: GET failure leaves local state untouched, next
poll retries.
- Docstring updated to reflect the actual behavior ("polls
transparently — keep polling and you'll see the flip").
- 4 new tests cover: QUEUED → completed via refresh; QUEUED →
failed via refresh; refresh keeps QUEUED when platform hasn't
resolved; refresh swallows network errors safely.
(2) Canvas side, AgentCommsPanel.tsx WS push handler:
- Listens for DELEGATION_SENT / DELEGATION_STATUS / DELEGATION_COMPLETE
/ DELEGATION_FAILED in addition to ACTIVITY_LOGGED.
- Each event's payload synthesized into an ActivityEntry shape
so toCommMessage's existing delegation branch maps it. Status
derived: STATUS uses payload.status, COMPLETE → "completed",
FAILED → "failed", SENT → "pending".
- The ACTIVITY_LOGGED branch keeps the "delegation" type accepted
as a no-op-today / future-proof path: if delegation handlers
are ever refactored to call LogActivity, this lights up
automatically without another canvas change.
Doesn't change: the docstring guidance ("queued → wait, don't bypass")
is now actually load-bearing because the refresh path will deliver
the eventual outcome. Without the refresh, the guidance was a trap.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two bugs that compounded into the "Director does the work itself" UX:
1. workspace/builtin_tools/delegation.py: _execute_delegation only
handled HTTP 200 in the response branch. When the peer's a2a-proxy
returned HTTP 202 + {queued: true} (single-SDK-session bottleneck
on the peer), the loop fell through. Two iterations later the
`if "error" in result` check tried to access an unbound `result`,
the goroutine ended quietly, and the delegation stayed at FAILED
with error="None". The LLM checking status saw "failed" + the
platform's "Delegation queued — target at capacity" log line in
chat context, concluded the peer was permanently unavailable, and
bypassed delegation to do the work itself.
Fix: explicit 202+queued branch. Adds DelegationStatus.QUEUED,
marks the local delegation as QUEUED, mirrors to the platform,
and returns cleanly without retrying. The retry loop is for
transient transport errors — queueing is a real ack, not a failure
to retry against (retrying would just re-queue the same task).
check_delegation_status docstring extended with explicit per-status
guidance: pending/in_progress → wait, queued → wait (peer busy on
prior task, reply WILL arrive), completed → use result, failed →
real error in error field; only fall back on failed, never queued.
2. canvas/src/components/tabs/chat/AgentCommsPanel.tsx: filter dropped
every delegation row because it whitelisted only a2a_send /
a2a_receive. activity_type='delegation' rows (written by the
platform's /delegate handler with method='delegate' or
'delegate_result') never reached toCommMessage. User saw "No
agent-to-agent communications yet" while 6+ delegations existed
in the DB.
Fix: include "delegation" in the both the initial filter and the
WS push filter, plus a delegation branch in toCommMessage that
maps the row as outbound (always — platform proxies on our behalf)
and uses summary as the primary text source.
Tests:
- 3 new Python tests cover the 202+queued path: status becomes
QUEUED not FAILED; no retry on queued (counted by URL match
against the A2A target since the mock is shared across all
AsyncClient calls); bare 202 without {queued:true} still
falls through to the existing retry-then-FAILED path.
- 3 new TS tests cover the delegation mapper: 'delegate' row
maps as outbound to target with summary text; queued
'delegate_result' preserves status='queued' (load-bearing for
the LLM's wait-vs-bypass decision); missing target_id returns
null instead of rendering a ghost.
Does NOT solve: the underlying single-SDK-session bottleneck that
causes peers to queue in the first place. Tracked as task #102
(parallel SDK sessions per workspace) — real architectural work.
This PR makes the runtime handle the queueing correctly so the LLM
doesn't bail out, and makes the delegations visible in Agent Comms
so operators can see what's happening.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Follow-up to #2110 (which generalised pruneStaleKeys to Map<string, T>).
Identified by the simplify reviewer on that PR as the only other
in-tree caller of the same shape: `for (const id of map.keys()) { if
(!liveIds.has(id)) map.delete(id); }`.
Net: -3 lines, one less hand-rolled GC loop. No behaviour change —
the helper does exactly what the inline block did.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Simplify pass on top of #2070 fix:
- Rename pruneStaleSubtreeIds → pruneStaleKeys, generalize to
Map<string, T> so the same shape can absorb other keyed-by-node-id
caches (ProvisioningTimeout.tsx tracking map is the obvious next
caller — left as a follow-up to keep this PR scoped).
- Trim the helper docstring to remove implementation-detail rot
(O(map_size), cadence claims). The ref-block comment carries the
rationale where it actually matters (at the call site).
- Add identity-preservation test: survivors must keep their original
Set reference. Guards against a future "rebuild instead of delete"
regression that would silently invalidate downstream === checks.
No behaviour change.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes#2070. The Map<rootId, Set<nodeId>> in useCanvasViewport.ts
accumulated entries indefinitely — adds on every successful auto-fit,
never deletes when a root left state.nodes (cascade delete or manual
remove). Operationally invisible until thousands of imports, but the
fix is cheap.
Adds pruneStaleSubtreeIds(map, liveNodeIds) — a pure helper exported
alongside the existing shouldFitGrowing helper, called at the top of
runFit before any read or write to the map. Bounds the map to "roots
present right now" instead of "every root ever auto-fitted in this
session." O(map_size) per fit; runs only at user-driven cadence.
Tests in __tests__/useCanvasViewport.test.ts cover the four cases:
delete-some / no-op / clear-all / never-add.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three files conflicted with staging changes that landed while this PR
sat open. Resolved each by combining both intents (not picking one side):
- a2a_proxy.go: keep the branch's idle-timeout signature
(workspaceID parameter + comment) AND apply staging's #1483 SSRF
defense-in-depth check at the top of dispatchA2A. Type-assert
h.broadcaster (now an EventEmitter interface per staging) back to
*Broadcaster for applyIdleTimeout's SubscribeSSE call; falls through
to no-op when the assertion fails (test-mock case).
- a2a_proxy_test.go: keep both new test suites — branch's
TestApplyIdleTimeout_* (3 cases for the idle-timeout helper) AND
staging's TestDispatchA2A_RejectsUnsafeURL (#1483 regression). Updated
the staging test's dispatchA2A call to pass the workspaceID arg
introduced by the branch's signature change.
- workspace_crud.go: combine both Delete-cleanup intents:
* Branch's cleanupCtx detachment (WithoutCancel + 30s) so canvas
hang-up doesn't cancel mid-Docker-call (the container-leak fix)
* Branch's stopAndRemove helper that skips RemoveVolume when Stop
fails (orphan sweeper handles)
* Staging's #1843 stopErrs aggregation so Stop failures bubble up
as 500 to the client (the EC2 orphan-instance prevention)
Both concerns satisfied: cleanup runs to completion past canvas
hangup AND failed Stop calls surface to caller.
Build clean, all platform tests pass.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Closes the canvas-side loop on #2054. Phases 1+2 plumbed
provision_timeout_ms from template manifest → workspace API →
canvas socket → node-data → ProvisioningTimeout resolver. The
template-hermes manifest declares provision_timeout_seconds: 720
(filed as a separate template-repo PR). With that flow live, the
canvas-side hardcoded RUNTIME_PROFILES.hermes entry is redundant.
Removed:
- RUNTIME_PROFILES.hermes (was 720000ms hardcoded in canvas/src/lib/runtimeProfiles.ts)
Doc updates:
- RUNTIME_PROFILES jsdoc explains the map is now empty by design —
new runtimes that need a non-default cold-boot threshold should
declare runtime_config.provision_timeout_seconds in their template
manifest, NOT add an entry here.
Tests updated (3):
- "returns hermes override when runtime = hermes" → "hermes returns
default — value moved server-side post-#2054 phase 3". Asserts
RUNTIME_PROFILES.hermes is undefined.
- The two server-override tests now compare against
DEFAULT_RUNTIME_PROFILE since hermes no longer has a profile entry.
19/19 pass locally. The end-state for hermes:
workspace-server reads template manifest at request time →
workspace API includes provision_timeout_ms: 720000 →
canvas hydrate populates node.data.provisionTimeoutMs →
ProvisioningTimeout resolver picks it up via overrides.
Same effective threshold (720s), now declarative and one-edit-point
per runtime.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
simplify-review note: the |/,-delimited node string is brittle if a
future string-typed field is added without sanitization. Document
which fields are user-typed (name — already sanitized) vs primitive
(id is UUID, runtime is a slug, provisionTimeoutMs is numeric) so
the next field-add doesn't accidentally introduce an injection
vector for the splitter.
Skipped (false-positive review finding): the agent flagged the
prop > runtime-profile order as inconsistent with the docstring,
but the docstring explicitly lists the prop at #2 (between node and
runtime-profile) — matches both the implementation AND the original
behavior pre-#2054 (the prop was 'timeoutMs ?? runtime-profile').
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Phase 1 of moving runtime UX knobs server-side. Builds the canvas
foundation: a workspace can carry its own provision_timeout_ms
(sourced server-side from a template manifest in a follow-up PR),
and ProvisioningTimeout's resolver respects it per-node.
Today the resolver had Props-level timeoutMs that applied to ALL
nodes — fine for tests but wrong for production where one batch
could mix runtimes (hermes 12-min cold boot alongside docker 2-min).
The runtime profile fallback already handles per-runtime defaults;
this PR adds the per-WORKSPACE override layer above that.
Resolution priority (most specific wins):
1. node.provisionTimeoutMs — server-declared per-workspace
override (this PR's new field)
2. timeoutMs prop — single-threshold test override
3. runtime profile in @/lib/runtimeProfiles
4. DEFAULT_RUNTIME_PROFILE
Changes:
- WorkspaceData (socket): add optional provision_timeout_ms
- WorkspaceNodeData: add optional provisionTimeoutMs
- canvas-topology hydrate: thread the field through to node.data
- ProvisioningTimeout: extend the serialized-string node iteration
to carry provisionTimeoutMs (4-field positional split); pass as
the second arg to provisionTimeoutForRuntime
- 3 new tests in ProvisioningTimeout.test.tsx covering hydrate
threading, null fall-through, and resolver priority
Phase 2 (separate PR, blocked on workspace-server template-config
loader): workspace-server reads provision_timeout_seconds from
template config.yaml at provision time, includes
provision_timeout_ms in the workspace API/socket response. Phase 3
(template-repo PR): template-hermes config.yaml declares
provision_timeout_seconds: 720; canvas RUNTIME_PROFILES.hermes
becomes redundant and can be removed.
19/19 tests pass (3 new + 16 existing).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bundle review of pieces 1/2/3 surfaced two critical issues plus a
handful of required + optional fixes. All addressed.
Critical:
1. Migration 043 was missing 'paused' and 'hibernated' from the
workspace_status enum. Both are real production statuses written
by workspace_restart.go (lines 283 and 406), introduced by
migration 029_workspace_hibernation. The original `USING
status::workspace_status` cast would have errored mid-transaction
on any production DB containing those values. Added both. Also
added `SET LOCAL lock_timeout = '5s'` so the migration aborts
instead of stalling the workspace fleet behind a slow SELECT.
2. The chat activity-feed window kept only 8 lines, and a single
multi-tool turn (Read 5 files + Grep + Bash + Edit + delegate)
easily flushed older context before the user could read it.
Extracted appendActivityLine to chat/activityLog.ts with a
20-line window AND consecutive-duplicate collapse (same tool
on the same target twice in a row is noise, not new progress).
5 unit tests pin the behavior.
Required:
3. The SDK wedge flag was sticky-only — a single transient
Control-request-timeout from a flaky network blip locked the
workspace into degraded for the whole process lifetime, even
when the next query() would have succeeded. Added
_clear_sdk_wedge_on_success(), called from _run_query's success
path. The next heartbeat after a working query reports
runtime_state empty and the platform recovers the workspace to
online without a manual restart. New regression test.
4. _report_tool_use now sets target_id = WORKSPACE_ID for self-
actions, matching the convention other self-logged activity
rows use. DB consumers joining on target_id see a well-defined
value instead of NULL.
Optional taken:
5. Tightened _WEDGE_ERROR_PATTERNS from "control request timeout"
to "control request timeout: initialize" — suffix-anchored so a
future SDK error on an in-flight tool-call control message
doesn't get misclassified as the unrecoverable post-init wedge.
6. Dropped the redundant "context canceled" substring fallback in
isUpstreamBusyError. errors.Is(err, context.Canceled) is the
typed check; the substring would also match healthy client-side
aborts, which we don't want classified as upstream-busy.
Verified: 1010 canvas tests + 64 Python tests + full Go suite pass;
migration applies cleanly on dev DB with all 8 enum values; reverse
migration restores TEXT.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two halves of the same UX win — the user wants to see what Claude is
doing while a chat reply is in flight instead of staring at "0s" for
minutes.
Workspace side (claude_sdk_executor.py):
- The executor's _run_query message loop already iterated the SDK
stream for AssistantMessage.TextBlock content. Now also detects
ToolUseBlock / ServerToolUseBlock entries (by class name, since
the conftest stub doesn't define them) and fires-and-forgets a
POST /workspaces/:id/activity row of type agent_log per tool use.
- _summarize_tool_use maps the common tools (Read, Write, Edit,
Bash, Glob, Grep, WebFetch, WebSearch, Task, TodoWrite) to a
one-line summary with the file path / pattern / command, falling
back to "🛠 <tool>(…)" for anything else. Truncated at 200 chars.
- Posts directly to /workspaces/:id/activity rather than going
through a2a_tools.report_activity, which would also push a
/registry/heartbeat current_task and double-log as a TASK_UPDATED
line in the same chat feed.
- All failures swallowed silently — telemetry must not break
the conversation.
Canvas side (ChatTab.tsx):
- The existing ACTIVITY_LOGGED handler streams a2a_send /
a2a_receive / task_update events into a sliding-window
activityLog state. Two issues fixed:
1. No `msg.workspace_id === workspaceId` filter — a sibling
workspace's a2a_send was leaking into the wrong chat
panel as "→ Delegating to X...". Added an early return.
2. No agent_log render branch. Added one that renders the
summary verbatim (the workspace already prefixed its
own emoji icon, so no double-icon).
- Existing 8-line sliding window keeps the UI scoped; older
progress lines naturally roll off as new ones arrive.
Result: when DD is delegating to Visual Designer + reading
config files + running Bash to lint, the spinner area shows:
📄 Read /configs/system-prompt.md
⚡ Bash: pnpm test
→ Delegating to Visual Designer...
← Visual Designer responded (47s)
instead of bare "0s · Processing with Claude Code..." for minutes.
63 Python tests + 58 canvas chat tests pass; tsc clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three required fixes from the bundle review of 391e1872:
1. workspace/a2a_client.py: substring `type_name in msg` could miss
the diagnostic prefix when an exception's message embedded a
different class name mid-string (e.g. `OSError("see ConnectionError
below")` → printed as plain msg, type lost). Switched to a
prefix-anchored check (`msg.startswith(f"{type_name}:")` etc.) so
the type label is always added when not already at the start of
the message.
2. workspace/a2a_tools.py: `activity_logs.error_detail` is unbounded
TEXT on the platform (handlers/activity.go does not validate
length). A buggy or hostile peer could stream arbitrarily large
error messages into the caller's activity log. Cap at 4096 chars
at the producer — comfortably above any real exception traceback,
well below an obvious-DoS threshold.
3. New regression test for JSON-RPC `code=0` — pins the
`code is not None` semantics so the code is preserved in the
detail rather than collapsing into the no-code path. Code=0 is
not valid per the spec, but a malformed peer can still emit it
and we want it visible for diagnosis.
Plus one optional taken: extracted the A2A-error → hint mapping into
canvas/src/components/tabs/chat/a2aErrorHint.ts. The two prior copies
(AgentCommsPanel.inferCauseHint + ActivityTab.inferA2AErrorHint) had
already drifted — Activity tab gained `not found`/`offline` cases the
chat panel never picked up, AgentCommsPanel handled empty-input
explicitly while Activity didn't. The shared module is the merged
superset, with 10 unit tests pinning each named pattern + the
"most specific first" ordering (Claude SDK wedge wins over generic
timeout).
Skipped (per analysis):
- Unicode-naive 120-char slice — Python str[:N] slices on code
points, not bytes. Safe.
- Nested [A2A_ERROR] confusion — non-issue per reviewer; outer
prefix winning still produces a structured render.
- MessagePreview + JsonBlock dual render on errors — intentional
drilldown; raw JSON is below the fold for operators who need it.
- console.warn dedup — refetches don't happen per-event so spam
risk is low.
- str(data)[:200] materialization — A2A response bodies aren't
typically MB-sized.
Verified: 1005 canvas tests pass (10 new hint tests); 10 Python
send_a2a_message tests pass (1 new for code=0); tsc clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Symptom: Activity tab and Agent Comms surfaced bare "[A2A_ERROR] "
(prefix + nothing) for failed delegations. Operator had no signal
to act on — no exception type, no target, no hint about what went
wrong, no next step. Fix is in three layers.
1. workspace/a2a_client.py — every error path now produces an
actionable detail string:
- except branch: some httpx exceptions (RemoteProtocolError,
ConnectionReset variants) stringify to "". Pre-fix the catch
was `f"{_A2A_ERROR_PREFIX}{e}"` → bare prefix. Now falls back
to `<TypeName> (no message — likely connection reset or silent
timeout)` and always appends `[target=<url>]` for traceability
in chained delegations.
- JSON-RPC error branch: previously dropped error.code on the
floor and printed "unknown" when message was missing. Now
surfaces both, including the well-defined "JSON-RPC error
with no message (code=N)" path.
- "neither result nor error" branch: pre-fix returned
str(payload) which the canvas rendered as a successful
response block. Now tagged as A2A_ERROR with a payload
snippet so downstream UI routes through the error path.
2. workspace/a2a_tools.py — tool_delegate_task now passes
error_detail (the stripped error message) through to the
activity-log POST. The platform's activity_logs.error_detail
column is the canvas's red error chip source; populating it
makes the failure visible in the row header without the user
having to expand into raw response_body JSON. The summary line
also gets a 120-char prefix of the cause so the collapsed row
reads "React Engineer failed: ConnectionResetError: ... [target=...]"
instead of "React Engineer failed".
3. canvas/src/components/tabs/ActivityTab.tsx — MessagePreview
now detects [A2A_ERROR]-prefixed bodies and renders a
structured error block (red chip, stripped detail, cause hint)
instead of the previous gray text-block that showed the literal
"[A2A_ERROR]" string. inferA2AErrorHint mirrors the patterns
from AgentCommsPanel.inferCauseHint so the same symptom reads
the same way in both surfaces (Claude SDK init wedge → restart
workspace; timeout → busy/stuck; connection-reset → transient
blip then check logs).
Tests: 9 send_a2a_message tests pass (including a new regression
test for the empty-stringifying-exception case that the user
reported); 995 canvas tests pass; tsc clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Reported symptom: canvas edges show "1 call · just now" between two
agents, but the Agent Comms tab for the source workspace renders
"No agent-to-agent communications yet" — even though
GET /workspaces/<id>/activity?source=agent&limit=50 returns a2a_send
+ a2a_receive rows.
Confirmed via curl that the API does return the rows the panel
should map. The panel's load handler was the suspect, but it had:
.catch(() => setLoading(false))
which swallowed every failure path — network errors, JSON parse,
ANY throw inside the .then body — without leaving a single trace in
the console. The panel just sat on its empty state and gave the user
zero signal to act on. (And by extension, gave us nothing to debug
remotely either.)
Two changes:
1. Wrap the per-row `toCommMessage` call in a try/catch so one
malformed activity row (unexpected request_body shape, etc.)
doesn't throw out of the for-loop and skip the
setMessages(msgs) line. Previously the panel would silently
drop the entire batch when ANY row failed to parse.
2. Replace the bare `.catch(() => setLoading(false))` with a
logging variant. Now a future "panel stuck empty" report comes
with `AgentCommsPanel: load activity failed <err>` or
`AgentCommsPanel: failed to map activity row {...}` in the
console — diagnosable instead of opaque.
Behavior on the happy path is unchanged (5 existing tests still
pass; tsc clean). This is purely defensive: it makes the failure
path visible so the next stuck-empty report can be root-caused
instead of guessed at.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bundle-level review caught an implicit coupling in useCanvasViewport
between two distinct fit effects:
- settle fit: 1200ms one-shot when provisioning transitions to zero
(deploy just finished — settle on the whole org once)
- tracking fit: 500ms debounced per molecule:fit-deploying-org event
(track the org's bounds as children land during the deploy)
Both effects shared a single autoFitTimerRef, so each one's
clearTimeout call could silently cancel the other's pending fit.
Today's behavior happened to land in the right order out of luck —
the tracking handler fires per-arrival during the deploy, then the
settle effect arms after the last child completes. But nothing in
the code enforces that ordering; a future refactor that, say,
fires the settle effect from the same event sequence as the
tracking timer (mid-deploy status flicker) would silently drop the
settle fit because the tracking timer's clearTimeout ran last.
Splitting into settleFitTimerRef + trackingFitTimerRef makes the
two effects fully independent. Cleanup clears both. Tests still pass
(995/995); the refactor is mechanical.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three review-driven fixes plus regression coverage for the bugs
landed in 176b703d / deedb5ef:
1. clearTimeout the prior reload handle before scheduling a new one in
both installFromSource and handleUninstall. Two installs within the
PLUGIN_RELOAD_DELAY_MS window (15s) used to queue two
loadInstalled() calls; the unmount cleanup only cleared the latest
handle, and the second reconciliation could overwrite a still-
correct optimistic state with a stale snapshot mid-restart.
2. Drop `setInstalledLoaded(true)` from the optimistic block. That
flag's contract is "the initial GET has succeeded at least once" —
it gates the auto-expand-registry effect. A user installing a
custom-source plugin BEFORE the initial fetch returned would flip
the gate prematurely, the auto-expand would never fire, and a
followup loadInstalled racing with the optimistic write could
overwrite our entry with [] mid-restart.
3. Don't force `supported_on_runtime: true` on the optimistic record.
The "inert on this runtime" badge in the row renders on the value
`=== false`. Forcing true would hide the badge for 15s if the user
installed a plugin that doesn't actually support the workspace's
runtime; the real value lands at refetch. Leaving the field
undefined keeps the badge neutral until reconciliation arrives.
Plus a behavioral test (SkillsTab.install.test.tsx) that asserts:
- the install POST URL contains the workspaceId (not "undefined")
- the row's "Install" button is replaced by the green "Installed"
tag synchronously after POST resolves, without advancing any
timer — locks in the optimistic-update contract so a future
refactor can't silently regress it.
995 canvas tests pass (2 new); tsc clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
After clicking Install, the button reverted from "Installing..." → "Install"
the moment the POST returned, then sat there for ~15s before the green
"Installed" tag appeared. The 15s gap is PLUGIN_RELOAD_DELAY_MS — we
delay the GET /workspaces/:id/plugins refetch to wait for the workspace
to restart (the listing handler returns [] while the container is
restarting because findRunningContainer comes up empty).
Uninstall already does optimistic local-state mutation (line 244 prior
to this commit) so the green tag → install button transition is
instant. Install was the inconsistent half — push the registry entry
into `installed` immediately after POST returns 200 and let the
delayed refetch reconcile.
The optimistic record uses the registry entry's metadata (name,
version, description, tags, runtimes, skills) and sets
supported_on_runtime=true. If reconciliation later disagrees (server
filter, install actually failed at the runtime layer), the refetch
overwrites the local record. Worst case is a brief 15s window where
we show "Installed" for a plugin that won't load — same window the
user previously experienced as "stuck on Install button" — but flipped
to the correct expected state.
Custom-source installs (github://, etc.) don't have a registry entry
to use, so they keep the old behavior of waiting for the refetch. Most
users install from the registry list in the UI.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
SkillsTab read \`data.id\` from its props and used the value to build
two API URLs:
POST /workspaces/\${data.id}/plugins
DELETE /workspaces/\${data.id}/plugins/\${pluginName}
But \`data\` is the React Flow node.data blob (WorkspaceNodeData) —
the workspace id lives on \`node.id\`, NOT on \`node.data\`. WorkspaceNodeData
extends \`Record<string, unknown>\`, which makes \`data.id\` type-check
silently as \`unknown\` instead of erroring. So every install/uninstall
hit \`/workspaces/undefined/plugins\`, the server's not-found path
returned 503 "workspace container not running" (misleading — the real
issue was the bogus URL), and the user got a confusing toast.
Every other tab in SidePanel takes \`workspaceId={selectedNodeId}\` as
an explicit prop. SkillsTab was the lone outlier, presumably because
"data has all the fields I need" is the obvious-looking shortcut that
TypeScript can't catch through the index-signature interface.
Fix: make \`workspaceId\` an explicit prop on SkillsTab, drop the
\`data.id\` reads, thread the prop from SidePanel like the other tabs.
Test fixture updated to pass it.
Verified: 993 canvas tests pass; tsc clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Independent code review surfaced two required documentation fixes and
one growth-correctness gap. All addressed here.
Auto-fit gate (useCanvasViewport):
The previous "subtree-grew-by-count" check missed the delete-then-add
case: subtree of 6 → delete one → 5 → a different child arrives → 6
again. A length-only comparison reads no growth and the fit is
skipped, leaving the new node off-screen. Switched to an id-set
membership snapshot so any brand-new id forces the fit even when the
count is unchanged.
The gate logic is now extracted as a pure exported function
`shouldFitGrowing(currentIds, prevIds, userPannedAt, lastAutoFitAt)`
so the regression-prone decision can be unit-tested in isolation
without standing up React Flow + DOM event refs. 8 cases cover:
first-fit, empty-prior, brand-new id, status-update with user pan,
no-pan-ever, pan-before-last-fit, delete-then-add same length, and
shrink-only with user pan.
Parser parity (dotenv.go + next.config.ts):
Existing-env semantics were undocumented in both parsers. Both now
explicitly note that an explicitly-set empty string (`KEY=` from the
parent shell) counts as "set" — the file value does NOT backfill —
matching the Go (os.LookupEnv) and Node (`process.env[k] !==
undefined`) primitives.
`export ` prefix uses a literal space; `export\tFOO=bar` is
intentionally rejected. Added the same comment in both parsers
to lock in this parity invariant since the commit message claims
"if one parser changes, the other has to."
Skipped (per analysis):
- Drag-pan respect for left-click drag-pan during deploy. The
growth-check safety net means any pan gets overridden on the
next arrival anyway, which is the desired behavior for the
"watch the org deploy" use case. After deploy completes, no
more fit-deploying-org events fire so drag-pan works freely.
- Map cleanup for lastFitSubtreeIdsRef. Per-tab session, UUID
keys, tiny entries — not worth the cleanup hook.
993 canvas tests pass (8 new); Go dotenv tests pass; tsc clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Symptom: org import zoomed to fit the parent + first child, then froze
at that framing while the remaining children kept materialising
off-screen. The user had to manually pan/zoom to see the new arrivals.
Two stacked bugs in useCanvasViewport's deploy-time auto-fit:
1. The user-pan-respect gate stamps userPannedAtRef on EVERY
pointerdown that lands inside .react-flow__pane. That fires for
ordinary clicks (deselect, click-near-a-card, modal-close-bubble
from the import dialog) — not just for actual pan gestures. One
accidental pre-import click was enough to lock out every fit for
the rest of the deploy. Wheel is the canonical unambiguous
pan/zoom signal; drop pointerdown.
2. Even with a real pan during deploy, when more children land the
org's bounds grow and the user has lost context — the new
arrivals are off-screen and the deploy is the primary thing they
want to watch right now. The guard had no growth awareness, so
one pan cancelled all follow-up fits unconditionally. Now we
track the subtree size at the last fit (per root), and if the
current subtree is larger we force the fit through regardless of
the user-pan timestamp. When the subtree size hasn't changed
(status updates on already-positioned nodes), the user-pan
respect still applies — so post-deploy exploration isn't
yanked back.
The Map keyed by root id supports back-to-back imports of different
orgs without one's growth count blocking the other's first fit.
985 canvas tests pass; tsc clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Deleting a parent on a wedged WS used to leave the child cards on
the canvas as orphaned roots until the user manually refreshed.
Why: Canvas.tsx and DetailsTab.tsx both called `removeNode(parentId)`
after `DELETE /workspaces/:id?confirm=true` returned 200. `removeNode`
deliberately re-parents children rather than cascading — it relies on
the per-descendant WORKSPACE_REMOVED WS events the platform emits as
part of the cascade to drop each child individually. When the WS is
unhealthy those events never arrive, so the local store keeps the
children alive (now re-parented to root since their actual parent is
gone).
Fix: new `removeSubtree(rootId)` action on the canvas store mirrors
the server-side cascade — drops the root + every descendant + every
incident edge in one atomic set(). Both delete call sites now use it.
The WS events still arrive when WS is healthy and become idempotent
no-ops because the nodes are already gone.
Why a new action instead of changing removeNode: removeNode's
re-parenting behavior is correct for non-cascading flows (drag-out,
manual node detach in the future). Adding a sibling action keeps
both call shapes available rather than forcing every caller to opt
out of cascade.
6 new unit tests cover root cascade, mid-level cascade, leaf
no-op-cascade, selection clearing across the subtree, selection
preservation outside the subtree, and edge cleanup.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two related fixes for the case where the canvas thinks workspaces are
stuck provisioning when they're actually online:
1. ProvisioningTimeout banners now gate on wsStatus === "connected".
While the WS is in connecting/disconnected state, the local
"provisioning" status reflects the last event received before the
drop — workspaces may have transitioned to online minutes ago. The
8m timeout was firing against frozen state and showing a wall of
yellow warnings on already-online workspaces.
2. Socket layer now starts a 10s rehydrate poll when the WS goes
unhealthy (onclose) and stops it on onopen/disconnect. The
reconnect attempts continue in parallel; whichever recovers first
wins. rehydrate()'s existing dedup gate prevents the open-time
rehydrate from racing with a fallback poll. Without this the
store could stay frozen for minutes while WS exponential backoff
chewed through retries.
Plus the previously-uncommitted TemplatePalette flushSync change so
the import modal unmounts synchronously before doImport runs (otherwise
React batches the close with the import's setState prefix and the
modal backdrop hides the spawn animation).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Single-themed bundle of fixes accumulated while polishing the canvas
chat / agent-comms / plugins / position flows. Each piece is small;
the connective tissue is "things observable from the canvas right
panel and the org-deploy flow that surprised real users".
UI / composer
- Legend: add close X + persisted-localStorage state + reopener
pill; default open for first-time users.
- SidePanel: rename "Skills" tab label → "Plugins" (single-line;
internal panelTab enum value, component name, and store keys
unchanged).
- SkillsTab: registry tri-state UI (loading / error / empty) with
actionable Retry button + 10s explicit fetch timeout. Handle
AbortSignal.timeout's DOMException by name (TimeoutError /
AbortError) — Chromium's "signal timed out" message wouldn't
match the prior naive /timeout/ regex. Reset mountedRef on every
mount: pre-existing StrictMode dev-mode bug where cleanup-only
`current = false` was never re-set, permanently wedging every
`if (mountedRef.current) setX(...)` guard and producing a
"Loading…" panel that never resolved on hard refresh.
- ChatTab: paste-image-from-clipboard via onPaste handler; unique
monotonic-counter filenames so same-second pastes don't collide
on name+size dedup. mime→ext map avoids `image/svg+xml`-style
raw extensions on synthesised filenames. Bypasses the
DataTransfer constructor so Safari < 14.1 / older Edge work.
- ChatTab: drop stuck error toast when the WS path already
delivered the agent reply but the HTTP path errored late
(sendingFromAPIRef gate now covers the .catch() handler).
- ChatTab: filter heartbeat-style internal self-messages from the
My Chat tab so historical rows with source_id=NULL don't
surface as user-typed input.
- Modal portals: OrgImportPreflightModal + MissingKeysModal
(ProviderPickerModal + AllKeysModal) now createPortal to
document.body and clamp max-h to 80vh. Escapes the ancestor
containing block (TemplatePalette's fixed+filtered sidebar
re-anchored descendants' position:fixed to itself, hiding
modals behind workspace cards). MissingKeysModal bumped to
z-[60] for stack ordering when both modals are open.
- OrgImportPreflightModal saveOne: ref-based microtask-safe
in-flight gate replaces the brittle "set startValue inside a
setState updater and read on the next line" pattern (React 18
doesn't guarantee functional updaters run synchronously; that
path strands `saving:true` and never calls createSecret). Same
useRef pattern guards SkillsTab.loadRegistry against concurrent
fires and Fast-Refresh-stranded promises; force=true parameter
on retry click bypasses the gate.
Agent comms
- AgentCommsPanel: derive UI-facing `flow` field instead of using
activity_type-derived direction. Self-logged a2a_receive rows
(source_id == workspace_id, what the agent runtime writes to log
its own outbound delegation replies) now correctly render as
OUTBOUND with → arrow + right-justified bubble. Previously they
rendered "← From Self" with Restart pointing at THIS workspace.
- AgentCommsPanel: error rows replace the unactionable
"X failed [A2A_ERROR]" body with banner + underlying-error
code-block + cause-hint (matched on Claude Code SDK init wedge,
deadline-exceeded, agent-thrown exception, empty-error) +
Restart [peer] / Open [peer] action buttons.
- AgentCommsPanel: render text bodies through ReactMarkdown +
remark-gfm so multi-part replies (tables, code) render properly.
Multi-part text extractor
- extractReplyText (live A2A response in ChatTab) and
extractResponseText (chat history loader in message-parser):
now COLLECT from every source — top-level parts, parts.root.text,
and artifacts — joined with "\n". Previous "first source wins"
silently dropped multi-part replies (Hermes summary+detail,
Claude Code long-form table). Tests cover joined-from-parts,
joined-from-artifacts, joined-from-both.
Position stability
- canvas-topology.buildNodesAndEdges: auto-rescue heuristic now
accepts currentParentSizes map; uses max(initial min, currently
grown) for the bbox check. Fixes "child jumps to weird location
after 30s" — the periodic socket health-check rehydrate
(silenceSec > 30) was rebuilding nodes from scratch, and the
rescue's reliance on grid-derived initial size false-flagged
children the user dragged into the user-grown area.
- canvas.hydrate: pass live measured dimensions from the existing
store into buildNodesAndEdges.
- socket.RehydrateDedup: pure exported helper class that gates
rehydrate calls. Two states — in-flight (in-flight Promise reused
by concurrent callers) + post-completion window (1.5s, returns
Promise.resolve()). Initialised with -Infinity so first call
always passes the gate. Wired into ReconnectingSocket.rehydrate.
A2A edges
- New A2AEdge custom React Flow edge component portals its label
out of the SVG layer via EdgeLabelRenderer so labels (a) render
above workspace cards instead of being hidden behind them and
(b) accept clicks. Click selects source + switches panel to
Activity, but only on a NEW selection (preserves current tab on
re-click of an already-selected source).
- buildA2AEdges output tagged type:"a2a"; edgeTypes wired in
Canvas.tsx.
Tests
- 14 new vitest cases across 4 files (964 → 978 passing):
OrgImportPreflightModal saveOne single-fire / double-click,
any-of rendering; AgentCommsPanel toCommMessage flow derivation
in all four shapes; canvas-topology rescue respects-grown /
rescues-genuine-drift / fallback-without-live-size; socket
RehydrateDedup gate behaviour; message-parser multi-part
response extraction.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extends the org-import env preflight so a template can declare an
alternative: satisfy ANY one member to pass. Motivated by the
Claude-family node case where either ANTHROPIC_API_KEY or
CLAUDE_CODE_OAUTH_TOKEN unlocks the agent — forcing both was wrong.
Server (workspace-server):
- New EnvRequirement union type with custom YAML + JSON
(un)marshaling. Accepts scalar (strict) or {any_of: [...]} in
both on-disk org.yaml and inline POST /org/import bodies.
- collectOrgEnv now returns []EnvRequirement. Dedups groups by
sorted-member signature. "Strict wins" pruning drops any-of
groups that mention a name already declared strictly (same
tier and cross-tier).
- Import preflight uses EnvRequirement.IsSatisfied — scalar =
exact match, group = any member present.
- Empty any_of: [] rejected at parse time (never-satisfiable).
- 14 handler tests (6 updated for the union shape, 8 new
covering any-of satisfaction, dedup, strict-dominates-group,
cross-tier pruning, invalid-member filtering, YAML round-trip,
and empty-any-of rejection).
Canvas:
- EnvRequirement = string | {any_of: string[]} with envReqMembers,
envReqSatisfied, envReqKey helpers.
- OrgImportPreflightModal renders strict rows and any-of groups
via a new AnyOfEnvGroup sub-component: "Configure any one"
banner, per-member input, ✓-satisfied indicator, and dimmed
siblings once any member is configured so the user can still
switch providers.
- TemplatePalette.OrgTemplate.required_env / recommended_env
retyped to EnvRequirement[]; passthrough to the modal
unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
## Problem
Two issues the external-workspace path was silently dropping:
1. `knownRuntimes` was a hardcoded Go map that drifted from
manifest.json — e.g. `gemini-cli` was in manifest but missing
from the Go allowlist, so any workspace provisioning with
runtime=gemini-cli got silently coerced to langgraph.
2. No end-to-end "bring your own compute" story. The canvas UI
had no way to pick runtime=external; the partial backend code
required the operator to already have a URL ready (chicken-and-
egg with the agent that doesn't exist yet), and no workspace_auth
_token was minted so the external agent couldn't authenticate its
register call.
## Change
### Runtime registry driven by manifest.json
- New `runtime_registry.go` reads `manifest.json` at service init.
Each `workspace_templates[].name` becomes a runtime identifier
(with the `-default` suffix stripped so `claude-code-default`
and `claude-code` resolve to the same runtime).
- `external` is always injected (no template repo exists for it).
- Falls back to a static map on manifest load failure so tests /
dev containers keep working.
- 5 new tests including a real-manifest sanity check.
### First-class external workspace flow
When `POST /workspaces` is called with `runtime: "external"` AND
no URL supplied:
1. Workspace row inserted with `status='awaiting_agent'`
(distinct from `provisioning` so canvas doesn't trip its
provisioning-timeout UX).
2. A workspace_auth_token is minted via `wsauth.IssueToken`.
3. Response body includes a `connection` object with:
- `workspace_id`, `platform_url`, `auth_token`
- `registry_endpoint`, `heartbeat_endpoint`
- `curl_register_template` — zero-dep one-shot register snippet
- `python_snippet` — full SDK setup w/ heartbeat loop,
paired with molecule-sdk-python PR #13's A2AServer
4. The platform URL is resolved from `EXTERNAL_PLATFORM_URL` env
(ops-configurable per tenant) or falls back to request headers.
The legacy `payload.External` + `payload.URL` path is preserved —
org-import and other callers that already have a URL still work.
### Canvas UI
- New "External agent (bring your own compute)" checkbox in
CreateWorkspaceDialog.
- When checked, template/model/hermes-provider fields are hidden
and the POST body includes `runtime: "external"`.
- New `ExternalConnectModal` component: shown once after create,
renders Python / curl / raw-fields tabs with copy-to-clipboard
buttons. Stays mounted as a sibling of the create dialog so the
token survives the create dialog unmount.
- `auth_token` is interpolated into the snippet client-side so the
copied block is truly ready to run — operator only has to fill
in their agent's public URL.
## Tests
- Go: 5 new runtime_registry tests (happy path, -default strip,
external always injected, missing file, malformed JSON, real
manifest sanity). All existing handler tests still pass.
- TypeScript: no type errors on my files; pre-existing
canvas-batch-partial-failure type drift is on main already and
tracked on the #2061 branch.
## Follow-ups (filed separately)
- Cut molecule-sdk-python v0.y to PyPI so the snippet can use
`pip install molecule-ai-sdk` instead of `git+main`.
- Add a `runtime: string` field per template in manifest.json so
one template can declare its runtime explicitly (instead of
deriving it from name conventions). Unblocks N-templates-per-
runtime (e.g. hermes-minimax, hermes-anthropic both runtime=hermes).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Builds on #2061. Three internally-cohesive sub-features; easiest to
read in order.
## 1. Org-level env preflight
Server
- `OrgTemplate` + `OrgWorkspace` gain `required_env: string[]` and
`recommended_env: string[]` YAML fields.
- `GET /org/templates` walks the tree and returns the tree-union
(deduped, sorted) of both. `collectOrgEnv` dedup prefers required
when the same key is declared at both tiers.
- `POST /org/import` preflights against `global_secrets` WHERE
`octet_length(encrypted_value) > 0` (empty-value rows used to be
counted as "configured" and the per-container preflight still
failed at start time). 412 Precondition Failed + `missing_env`
list when required keys are absent. `force=true` bypasses with
an audit log line. DB lookup failure now returns 500 (was:
silent fall-through that defeated the guard). Env-var NAMES
validated against `^[A-Z][A-Z0-9_]{0,127}$` so a malicious
template can't ship pathological names into the UI or DB.
Canvas
- New `OrgImportPreflightModal`: red "Required" section (blocking)
and yellow "Recommended" section (non-blocking, import stays
enabled, shows live missing-count next to the Import button).
- Per-key password input → `PUT /settings/secrets` → strike-through
on save. Functional `setDrafts` throughout (no stale-closure
clobbers on rapid successive saves). `useEffect` seed keyed on a
sorted-join string signature so a parent re-render with a new
array identity doesn't clobber typed inputs.
- `TemplatePalette.handleImport` branches: zero env declarations →
straight to import; any declarations → fetch configured global
secret keys, open the modal.
Tests (Go): `TestCollectOrgEnv_*` (5) cover union-across-levels,
required-wins-over-recommended (including same-struct), dedup,
empty, invalid-name rejection.
## 2. EmptyState parity with TemplatePalette
The "Deploy your first agent" grid used to call `POST /workspaces`
with no preflight while the sidebar palette ran
`checkDeploySecrets` + `MissingKeysModal` first. Same template
deployed two different ways → first-run users saw containers boot
in `failed` state without guidance. Now both surfaces share one
preflight + modal handshake.
EmptyState's previous `interface Template` dropped `runtime`,
`models`, and `required_env` — silently discarding exactly the
fields the preflight needs. `Template` now lives in
`deploy-preflight.ts` and is imported from there by both surfaces.
## 3. useTemplateDeploy hook
With the preflight + modal wiring now duplicated across
EmptyState + TemplatePalette + (going forward) any third surface,
extracted the pattern into `canvas/src/hooks/useTemplateDeploy.tsx`:
const { deploy, deploying, error, modal } = useTemplateDeploy({
canvasCoords: ..., // optional, default random
onDeployed: (id) => ...,
});
Closes three drift surfaces that the duplication had created:
- `resolveRuntime` id→runtime fallback table (moved to
`deploy-preflight.ts`). EmptyState had a narrower fallback that
would have silently disagreed with the palette on any future id
needing a non-identity mapping.
- `checkDeploySecrets` call signature. One owner.
- `MissingKeysModal` JSX wiring. One owner.
Narrow try/catch around `checkDeploySecrets` so a preflight network
failure clears `deploying` and surfaces via `setError` instead of
stranding the button forever. `modal: ReactNode` (not a
`renderModal()` function) — the previous memoization bought
nothing since consumers called it inline every render. Named
`MissingKeysInfo` interface for the state shape.
## 4. Viewport auto-fit user-pan gate fix
During org deploy the canvas was meant to pan+zoom to follow each
arriving workspace (`molecule:fit-deploying-org` event → debounced
fitView). In practice the fit stayed stuck on wherever the first
fit landed.
Root cause: React Flow v12 fires `onMoveEnd` with a truthy `event`
at the END of a programmatic `fitView` animation. The original
"respect-user-pan" gate stamped `userPannedAtRef` in `onMoveEnd`,
so our own fit completing looked like a user pan, and every
subsequent auto-fit short-circuited for the rest of the deploy.
Fix: stop trusting `onMoveEnd` for user-intent detection. Register
explicit `wheel` + `pointerdown` listeners on `document` with
capture phase and `target.closest('.react-flow__pane')` filter.
Capture-phase immunity to `stopPropagation`; pane-filter rejects
toolbar / modal / side-panel clicks (the old `window` fallback
caught those). `onMoveEnd` simplified to only drive the debounced
viewport save.
Also: fit event dispatched on root arrivals (not just children),
so the canvas centers on the just-landed root immediately instead
of waiting ~2s for the first child. Animation 600ms → 400ms so
successive per-arrival fits don't pile up visually. End-state fit
stays at 1200ms — intentional asymmetry ("settling" vs
"tracking"), documented in code.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Merge origin/staging into fix/canvas-multilevel-layout-ux. 18 files
auto-merged (mostly canvas/tabs/chat and workspace-server handlers
the earlier DIRTY marker was stale relative to current staging).
- Fix 7 test failures surfaced by the merge:
1. Canvas.pan-to-node.test.tsx — mockGetIntersectingNodes was
inferred as vi.fn(() => never[]); mockReturnValueOnce of a node
object failed type check. Explicit return-type annotation.
2. Canvas.pan-to-node.test.tsx + Canvas.a11y.test.tsx — Canvas.tsx
reads deletingIds.size (new multilevel-layout state). Both mock
stores lacked deletingIds; added new Set<string>() to each.
3. canvas-batch-partial-failure.test.ts — makeWS() built a wire-
format WorkspaceData (snake_case, with x/y/uptime_seconds). The
store's node.data is now WorkspaceNodeData (camelCase, no wire-
only fields). Rewrote makeWS to produce WorkspaceNodeData and
updated 5 call-site casts. No assertions changed.
4. ConfigTab.hermes.test.tsx — two tests pinned pre-#2061 behavior
that the PR intentionally inverts:
a. "shows hermes-specific info banner" — RUNTIMES_WITH_OWN_CONFIG
now contains only {"external"}, so the banner is no longer
shown for hermes. Inverted assertion: now pins ABSENCE of
the banner, with a comment noting the inversion.
b. "config.yaml runtime wins over DB" — priority reversed:
DB is now authoritative so the tier-on-node badge matches
the form. Inverted scenario: DB=hermes + yaml=crewai →
form shows hermes. Switched test's DB runtime off langgraph
because the dropdown collapses langgraph into an empty-
valued "default" option that would hide the win signal.
- No production code changed — this commit is staging merge + test
realignment only. 953/953 canvas tests pass. tsc --noEmit clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Session's accumulated UX work across frontend and platform. Reviewable
in four logical sections — diff is large but internally cohesive
(each section fixes a gap the next one depends on).
## Chat attachments — user ↔ agent file round trip
- New POST /workspaces/:id/chat/uploads (multipart, 50 MB total /
25 MB per file, UUID-prefixed storage under
/workspace/.molecule/chat-uploads/).
- New GET /workspaces/:id/chat/download with RFC 6266 filename
escaping and binary-safe io.CopyN streaming.
- Canvas: drag-and-drop onto chat pane, pending-file pills,
per-message attachment chips with fetch+blob download (anchor
navigation can't carry auth headers).
- A2A flow carries FileParts end-to-end; hermes template executor
now consumes attachments via platform helpers.
## Platform attachment helpers (workspace/executor_helpers.py)
Every runtime's executor routes through the same helpers so future
runtimes inherit attachment awareness for free:
- extract_attached_files — resolve workspace:/file:///bare URIs,
reject traversal, skip non-existent.
- build_user_content_with_files — manifest for non-image files,
multi-modal list (text + image_url) for images. Respects
MOLECULE_DISABLE_IMAGE_INLINING for providers whose vision
adapter hangs on base64 payloads (MiniMax M2.7).
- collect_outbound_files — scans agent reply for /workspace/...
paths, stages each into chat-uploads/ (download endpoint
whitelist), emits as FileParts in the A2A response.
- ensure_workspace_writable — called at molecule-runtime startup
so non-root agents can write /workspace without each template
having to chmod in its Dockerfile.
Hermes template executor + langgraph (a2a_executor.py) + claude-code
(claude_sdk_executor.py) all adopt the helpers.
## Model selection & related platform fixes
- PUT /workspaces/:id/model — was 404'ing, so canvas "Save"
silently lost the model choice. Stores into workspace_secrets
(MODEL_PROVIDER), auto-restarts via RestartByID.
- applyRuntimeModelEnv falls back to envVars["MODEL_PROVIDER"]
so Restart propagates the stored model to HERMES_DEFAULT_MODEL
without needing the caller to rehydrate payload.Model.
- ConfigTab Tier dropdown now reads from workspaces row, not the
(stale) config.yaml — fixes "badge shows T3, form shows T2".
## ChatTab & WebSocket UX fixes
- Send button no longer locks after a dropped TASK_COMPLETE —
`sending` no longer initializes from data.currentTask.
- A2A POST timeout 15 s → 120 s. LLM turns routinely exceed 15 s;
the previous default aborted fetches while the server was still
replying, producing "agent may be unreachable" on success.
- socket.ts: disposed flag + reconnectTimer cancellation + handler
detachment fix zombie-WebSocket in React StrictMode.
- Hermes Config tab: RUNTIMES_WITH_OWN_CONFIG drops 'hermes' —
the adaptor's purpose IS the form, banner was contradictory.
- workspace_provision.go auto-recovery: try <runtime>-default AND
bare <runtime> for template path (hermes lives at the bare name).
## Org deploy/delete animation (theme-ready CSS)
- styles/theme-tokens.css — design tokens (durations, easings,
colors). Light theme overrides by setting only the deltas.
- styles/org-deploy.css — animation classes + keyframes, every
value references a token. prefers-reduced-motion respected.
- Canvas projects node.draggable=false onto locked workspaces
(deploying children AND actively-deleting ids) — RF's
authoritative drag lock; useDragHandlers retains a belt-and-
braces check.
- Organ cancel button (red pulse pill on root during deploy)
cascades via existing DELETE /workspaces/:id?confirm=true.
- Auto fit-view after each arrival, debounced 500 ms so rapid
sibling arrivals coalesce into one fit (previous per-event
fit made the viewport lurch continuously).
- Auto-fit respects user-pan — onMoveEnd stamps a user-pan
timestamp only when event !== null (ignores programmatic
fitView) so auto-fits don't self-cancel.
- deletingIds store slice + useOrgDeployState merge gives the
delete flow the same dim + non-draggable treatment as deploy.
- Platform-level classNames.ts shared by canvas-events +
useCanvasViewport (DRY'd 3 copies of split/filter/join).
## Server payload change
- org_import.go WORKSPACE_PROVISIONING broadcast now includes
parent_id + parent-RELATIVE x/y (slotX/slotY) so the canvas
renders the child at the right parent-nested slot without doing
any absolute-position walk. createWorkspaceTree signature gains
relX, relY alongside absX, absY; both call sites updated.
## Tests
- workspace/tests/test_executor_helpers.py — 11 new cases
covering URI resolution (including traversal rejection),
attached-file extraction (both Part shapes), manifest-only
vs multi-modal content, large-image skip, outbound staging,
dedup, and ensure_workspace_writable (chmod 777 + non-root
tolerance).
- workspace-server chat_files_test.go — upload validation,
Content-Disposition escaping, filename sanitisation.
- workspace-server secrets_test.go — SetModel upsert, empty
clears, invalid UUID rejection.
- tests/e2e/test_chat_attachments_e2e.sh — round-trip against
a live hermes workspace.
- tests/e2e/test_chat_attachments_multiruntime_e2e.sh — static
plumbing check + round-trip across hermes/langgraph/claude-code.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Marketing-lead agent's rename pass updated the "renders all three plans"
test (lines 56-57) but missed lines 77, 94, 114, 132, 143, 158 which still
referenced the pre-rename "Upgrade to Starter" / "Upgrade to Pro" button
names. Canvas (Next.js) build failed with getByRole timeout because the
component now says "Upgrade to Team" / "Upgrade to Growth".
Internal PlanId tuple ("free" | "starter" | "pro") and startCheckout(planId)
call are unchanged — only the user-facing button labels shifted, so
assertions like startCheckout("pro", "acme") still match the server-side API.
Verified locally: 9/9 PricingTable tests pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Schema-driven ChannelsTab renders no inputs when config_schema is
absent — the test's bare {type, display_name} mock mismatched the
real API shape and every getByLabelText("Bot Token") failed.
Mock now mirrors GET /channels/adapters with the Telegram schema
(bot_token password + chat_id text) so the a11y assertions run
against the actual rendered form.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Lark adapter was already implemented in Go (lark.go — outbound Custom Bot
webhook + inbound Event Subscriptions with constant-time token verify),
but the Canvas connect-form hardcoded a Telegram-shaped pair of inputs
(bot_token + chat_id). Selecting "Lark / Feishu" from the dropdown
silently sent the wrong field names — there was no way to enter a
webhook URL.
Fix: move form shape to the server.
- Add `ConfigField` struct + `ConfigSchema()` method to the
`ChannelAdapter` interface. Each adapter declares its own fields with
label/type/required/sensitive/placeholder/help.
- Implement per-adapter schemas:
- Lark: webhook_url (required+sensitive) + verify_token (optional+sensitive)
- Slack: bot_token/channel_id/webhook_url/username/icon_emoji
- Discord: webhook_url + optional public_key
- Telegram: bot_token + chat_id (unchanged UX, keeps Detect Chats)
- Change `ListAdapters()` to return `[]AdapterInfo` with config_schema
inline. Sorted deterministically by display name so UI ordering is
stable across Go's random map iteration.
- Update the 3 existing `ListAdapters` test sites to struct access.
Canvas (`ChannelsTab.tsx`):
- Replace the two hardcoded bot_token/chat_id inputs with a single
schema-driven `SchemaField` component. Renders one input per field in
the order the adapter returns them.
- Form state becomes `formValues: Record<string,string>` keyed by
`ConfigField.key`. Values reset on platform-switch so stale
Telegram credentials can't leak into a new Lark channel.
- "Detect Chats" stays but only renders for platforms in
`SUPPORTS_DETECT_CHATS` (Telegram only — the only provider with
getUpdates).
- Only schema-known keys are posted in `config`, scrubbing any stale
values from previous platform selections.
Regression tests:
- `TestLark_ConfigSchema` locks in the 2-field Lark contract with the
required/sensitive flags correctly set.
- `TestListAdapters_IncludesLark` confirms registry wiring + schema
survives round-trip through ListAdapters.
Known pre-existing `TestStripPluginMarkers_AwkScript` failure in
internal/handlers is unrelated to this change (verified via stash+test
on clean staging).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Preparation for a "hundreds of runtimes" plugin ecosystem. Keeping the
runtime-specific UX knobs in-line inside ProvisioningTimeout scales badly
— every new runtime would require editing a component, not just adding a
table entry. Other components (create-workspace dialog, workspace card
tooltips, etc.) will want the same runtime metadata.
Changes:
- New file `canvas/src/lib/runtimeProfiles.ts` owns:
* `RuntimeProfile` type — structural shape, every field optional so
new runtimes can partially-fill without breaking consumers.
* `DEFAULT_RUNTIME_PROFILE` — 2-min default floor (docker-fast).
* `RUNTIME_PROFILES` — named overrides (currently: hermes 12 min).
* `WorkspaceRuntimeOverrides` — interface for server-provided
per-workspace overrides, so operators can tune via template
manifest / workspace metadata without a canvas release.
* `getRuntimeProfile()` — resolver with
overrides → profile → default priority.
* `provisionTimeoutForRuntime()` — convenience wrapper.
- `ProvisioningTimeout.tsx` now delegates to the profile module.
`DEFAULT_PROVISION_TIMEOUT_MS` re-exported for legacy test importers.
- Tests: 16/16 (up from 9 before the first fix). Adds pinning for:
* overrides > profile > default priority chain
* "every entry in RUNTIME_PROFILES resolves to a number" contract
* backward-compat export
Adding a new slow runtime is now one table entry in
`canvas/src/lib/runtimeProfiles.ts` with a mandatory `WHY` comment.
Moving to server-driven profiles later is a ~10-line change (the
resolver already threads WorkspaceRuntimeOverrides through).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Hermes workspaces cold-boot in 8-13 min (ripgrep + ffmpeg + node22 +
hermes-agent source build + Playwright + Chromium ~300MB). The canvas's
2-min hardcoded "Provisioning Timeout" warning fired at ~2min and told
users their workspace was "stuck" while it was still mid-install. Users
hit Retry, triggering fresh cold boots and cancelling healthy workspaces.
User-facing symptom (reported 2026-04-24 18:35Z): hermes workspace showed
"has been provisioning for 3m 15s — it may have encountered an issue"
with Retry + Cancel buttons, while the EC2 was installing node_modules.
Fix:
- Keep DEFAULT_PROVISION_TIMEOUT_MS = 120_000 (2min) — correct for fast
docker runtimes (claude-code, langgraph, crewai) where cold boot is
30-90s.
- Add RUNTIME_TIMEOUT_OVERRIDES_MS = { hermes: 720_000 } (12min).
Aligns with tests/e2e/test_staging_full_saas.sh's
PROVISION_TIMEOUT_SECS=900 (15min) so UI warns shortly before the
backend itself gives up.
- New timeoutForRuntime() resolves the base; per-node lookup in the
check-timeouts interval so a mixed batch (1 hermes + 2 langgraph) uses
the right threshold for each.
- timeoutMs prop is now optional. Undefined → per-runtime lookup; a
number → forces a single threshold for every workspace (tests use this
for deterministic behavior).
Tests: 4 new cases pinning the runtime-aware resolution, including a
guard that catches future regressions that would weaken hermes's budget.
Existing tests unchanged (they import DEFAULT_PROVISION_TIMEOUT_MS which
still exports 120_000).
13/13 pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
EmbeddedTeam was defined in WorkspaceNode.tsx but had no call site —
TeamMemberChip (which is called directly) covers the same rendering
responsibility. The function was stranded after a prior refactor and
was flagged by github-code-quality on PR #1989 (merged 2026-04-24T14:09Z
without this cleanup because the token died before push).
Removes 25 lines of dead code. MAX_NESTING_DEPTH is kept — it is used
by TeamMemberChip at line 498.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
TeamMemberChip used MAX_NESTING_DEPTH to cap recursive sub-agent
rendering at depth 3, but the constant was never declared — causing
a TypeScript build error ('Cannot find name MAX_NESTING_DEPTH') that
blocked Canvas CI on PR #1989.
Add the constant above EmbeddedTeam with a doc comment explaining its
purpose (guards against circular parentId cycles + readability cap).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
CI failure: "Cannot find name 'useMemo'" at line 363.
useMemo was called but not imported — likely dropped during refactor.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Five tightly-related fixes surfaced while stress-testing org-template
imports (Legal Team, Molecule Company, etc.) on a running control plane:
1) Org import was silently failing — INSERT wrote `collapsed` into the
`workspaces` table but that column lives on `canvas_layouts`
(005_canvas_layouts.sql). Every import returned 207 with 0 rows
created, which `api.post` treated as success → green "Imported"
toast + empty canvas. Moved the write to canvas_layouts; updated
the workspace_crud PATCH path to UPSERT there too; refreshed the
test mock. Added a client-side assertion that throws on
2xx-with-`error`-body so future partial-failures surface a red
toast rather than lying about success.
2) Multi-level nested layout was collision-prone: children that were
themselves parents (CTO → Dev Lead → 6 engineers) got the same
leaf-sized grid slot as leaf siblings and clipped into each other.
Added post-order `sizeOfSubtree` + sibling-size-aware
`childSlotInGrid` on both the Go server and the TS client (kept in
sync). `buildNodesAndEdges` now uses subtree sizes for both parent
dimensions and the rescue heuristic. `setCollapsed` on expand now
reads each child's actual rendered width/height instead of the
leaf-count formula — a regression test covers the CTO/Dev Lead
scenario.
3) Provisioning-timeout banner was unusable during large imports: a
30-workspace tree triggered 27 simultaneous "stuck" warnings 2
minutes in (server paces + provision concurrency = 3 guarantee tail
items legitimately wait longer). Scaled threshold with concurrent
count (base + 45s per queue slot beyond concurrency) and added a
Dismiss (×) button per banner.
4) Auto pan-and-zoom on org ready: after the last workspace flips out
of `provisioning`, canvas now fitView's with a 1.2s animation,
0.25 padding, `maxZoom: 0.8` and `minZoom: 0.25`. Without the zoom
caps fitView was hitting the component's maxZoom=2 on small trees
and zooming in instead of out.
5) Toolbar was visually busy: `+ N sub` count wrapped onto a second
row on narrow viewports; status dot and workspace total were in
separate border-delimited cells. Merged into one segment with
`whitespace-nowrap`; A2A / Audit / Search / Help collapsed to
icon-only 28px buttons with tooltip + aria-label (Figma/Linear
pattern). Stop All / Restart Pending keep text — they're urgent.
Also:
- `api.{get,post,...}` accept an optional `{ timeoutMs }` so callers
that hit intentionally-slow endpoints (org import paces 2s between
siblings) don't trip the 15s default and report false aborts.
- `WorkspaceNode` clamps role text to 2 lines so verbose descriptions
don't unboundedly grow card height and break the grid.
- `PARENT_HEADER_PADDING` bumped 44→130 to clear name + runtime +
2-line role + the currentTask banner that appears during the
initial-prompt phase.
Tests: 930 canvas tests + full Go handler suite pass. Added
regressions for (i) 207 partial-success surfacing as throw, and
(ii) setCollapsed sizing with nested-parent children.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
AuthGate now skips session fetch for /cp/auth/* paths, and
redirectToLogin guards against re-setting window.location when
already on an auth path. Both guards had no test coverage —
a future refactor could silently reintroduce the redirect loop.
Added:
- AuthGate.test.tsx: 2 cases covering /cp/auth/login and
/cp/auth/signup path skipping (no fetchSession call, no
redirectToLogin call, children rendered)
- auth.test.ts: 2 cases covering redirectToLogin early return
for /cp/auth/login and /cp/auth/signup paths
Fixes: Molecule-AI/molecule-core#1541
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
