molecule-core/docs/edit-history/2026-04-08.md

2026-04-08 Session

Summary

Fixed ChatTab agent reachability, added conversation history to all A2A adapters, added current_task heartbeat reporting, fixed WORKSPACE_PROVISIONING for restarts, fixed Config tab runtime dropdown, and improved config save/restart UX.

Changes

ChatTab — Agent Reachability Fix

• Problem: ChatTab called GET /registry/discover/:id without X-Workspace-ID header → 400 error → "Agent not available" even though agent was online
• Fix: Derived reachability from data.status (online/degraded) instead of network call. Messages are proxied through POST /workspaces/:id/a2a so browser never needs the agent's internal URL.
• Files: canvas/src/components/tabs/ChatTab.tsx

Conversation History

• ChatTab now sends last 20 messages via params.metadata.history in A2A message/send
• a2a_executor.py: New _extract_history() function extracts history from request metadata
• LangGraph/DeepAgents: History prepended as ("human"/"ai", text) tuples
• CrewAI/AutoGen: History prepended as text prefix in task description
• Files: ChatTab.tsx, a2a_executor.py, all adapter files

Current Task Heartbeat

• New shared set_current_task(heartbeat, task) function in a2a_executor.py
• All 5 adapters now set current_task during execution (truncated to 60 chars)
• Task cleared in finally block after execution completes
• Heartbeat passed from AdapterConfig through create_executor() in all adapters
• Files: a2a_executor.py, langgraph/adapter.py, deepagents/adapter.py, crewai/adapter.py, autogen/adapter.py, openclaw/adapter.py

WORKSPACE_PROVISIONING for Restarts

• Problem: applyEvent WORKSPACE_PROVISIONING only created new nodes, silently ignored restarts of existing nodes → UI didn't show "starting" state
• Fix: Added else branch that sets existing node to status: "provisioning", clears needsRestart and currentTask
• Files: canvas/src/store/canvas.ts

Config Tab Improvements

• Runtime dropdown: Removed invalid options (Codex, Ollama). Now shows only available adapters: LangGraph, Claude Code, CrewAI, AutoGen, DeepAgents, OpenClaw
• Save & Restart: Config save now auto-restarts workspace so changes take effect immediately. "Save" button also available for save-only (sets needsRestart banner)
• Secrets: Removed needsRestart: true from secrets save/delete since platform already auto-restarts
• Retry→Restart: Chat error banner button changed from no-op "Retry" to functional "Restart" with confirmation dialog
• Files: canvas/src/components/tabs/ConfigTab.tsx, ChatTab.tsx

Tests

• 8 new Python tests (15 total in test_a2a_executor.py, 80 total):
  • _extract_history: 5 tests (basic, empty, None, malformed, non-list)
  • History prepend in executor: 1 test
  • set_current_task: 2 tests (update + None heartbeat)
• 1 updated Canvas test: WORKSPACE_PROVISIONING updates existing node status on restart
• All existing tests updated ("user" → "human" role format, metadata in mock context)

Code Review Fixes

• PEP 8 spacing in all set_current_task() calls
• OpenClaw set_current_task("") moved into finally block
• _extract_history guards against non-dict entries in history list

Merged PR #1: Workspace Awareness Integration

• Platform assigns deterministic awareness_namespace (workspace:<id>) per workspace
• AWARENESS_URL and AWARENESS_NAMESPACE injected into containers during provisioning (only when AWARENESS_URL env var is set on the platform)
• commit_memory / search_memory tools route through awareness when configured, fall back to platform memory API
• New migration 010_workspace_awareness.sql adds awareness_namespace column to workspaces
• agent.py: Anthropic/OpenAI base URL support via ANTHROPIC_BASE_URL / OPENAI_BASE_URL env vars
• test_sandbox.py: asyncio.get_event_loop() → asyncio.run() for Python 3.13 compat
• New files: workspace/tools/awareness_client.py, workspace/tests/test_memory.py, workspace/tests/test_agent_base_urls.py
• Files: workspace-server/internal/handlers/workspace.go, workspace-server/internal/models/workspace.go, workspace-server/internal/provisioner/provisioner.go, workspace-server/migrations/010_workspace_awareness.sql, workspace/agent.py, workspace/main.py, workspace/tools/memory.py, workspace/tools/awareness_client.py

Restart Runtime Detection + Template Fallback

• Problem: Changing runtime via Config tab (e.g. langgraph → claude-code) didn't take effect on restart — provisioner used the old image because it only read runtime from the template dir, not the container's config volume
• Fix: Restart handler reads runtime from the running container via ExecRead (docker exec cat) BEFORE stopping it. Falls back to this value when no template provides a runtime.
• Template auto-apply: When a runtime has a default template (e.g. claude-code-default/), it's automatically applied on restart — copies CLAUDE.md, .claude/settings.json, etc. into the container
• Replaced ReadFileFromVolume (temp Alpine container, slow) with ExecRead (exec in existing container, instant)
• Files: workspace-server/internal/handlers/workspace.go, workspace-server/internal/provisioner/provisioner.go

MCP Memory Tools for CLI Runtimes

• Added commit_memory and recall_memory to a2a_mcp_server.py — now ALL runtimes (including Claude Code) can persist and recall memories via platform API
• Updated workspace-configs-templates/claude-code-default/CLAUDE.md with memory usage guidelines (recall at conversation start, commit after interactions)
• 7 unit tests in test_mcp_memory.py + 16 new E2E checks for memory CRUD, scope filtering, cross-workspace isolation

Comprehensive Test Suite

• registry/access_test.go: 10 tests for CanCommunicate (siblings, parent-child, root, denied, grandchild)
• handlers_extended_test.go: 14 tests for Delete, Update, Restart, Secrets, Discover, Peers, CheckAccess, Bundle, Config
• test_cli_executor.py: 14 tests for CLI command building, session resume, model flags, timeout
• test_plugins.py: 9 tests for plugin loading (rules, skills, prompts)
• test_comprehensive_e2e.sh: 68 checks covering ALL platform endpoints including runtime assignment and memory

UI Cleanup

• Removed 3 redundant task notifications from SidePanel/ChatTab (kept only the amber banner below tabs)
• PM system prompt updated for fully autonomous delegation (no more "Shall I delegate?")

Runtime Persisted in Database (migration 011)

• Root cause: runtime was only in config.yaml inside Docker volumes — fragile detection via ExecRead/ReadFromVolume failed when containers were dead
• Fix: Added runtime column to workspaces table. Stored at creation, read on restart with simple SELECT
• Fixed 6 broken paths: Restart, RestartByID, Create, Update (PATCH), Bundle import, ConfigTab
• Removed ExecRead/ReadFromVolume workarounds entirely

Auto-Memory for CLI Agents

• cli_executor.py: auto-recalls memories on first message (no session), auto-commits summary after each response
• Memories persist via platform API, survive container restarts
• Fixed memory pollution: saves original input, not memory-injected version

MCP Memory Tools

• Added commit_memory and recall_memory to a2a_mcp_server.py — all runtimes can persist/recall memories
• Updated claude-code-default/CLAUDE.md with memory guidelines

Real-Time Task Status on Canvas

• set_current_task pushes heartbeat immediately when setting a task (not just on 30s loop)
• Clearing deferred to next heartbeat cycle — keeps task visible for quick A2A responses
• Team leads now show task banners during delegation

Auth & Session Fixes

• CLI executor clears session_id on auth errors (prevents poisoned session resume)
• FilesTab: deduplicated tree keys with path:type (.claude dir + file collision)

UX Improvements

• Chat tab is now first and default tab (was Details)
• Rate limit increased from 100 to 600 req/min (15 workspaces overwhelmed the default)
• Merged PR #3: Awareness memory dashboard embedded as iframe in Memory tab

CI Fixes

• Updated handler tests for runtime column (INSERT 7 args, SELECT includes runtime)

Build Fixes

• workspace/Dockerfile: Added COPY policies/ ./policies/
• workspace/requirements.txt: Added langchain-core to base deps
• adapters/crewai/adapter.py: Fixed _langchain_to_crewai docstring

Container Health Detection & Auto-Restart

• Problem: When Docker Desktop crashes, containers die but platform still thinks workspaces are "online" for up to 60s (Redis TTL). A2A proxy returns errors, terminal fails, discovery returns stale URLs.
• Three-layer fix:
  1. Reactive: A2A proxy checks provisioner.IsRunning() on connection error → marks offline, clears Redis, triggers restart. Returns 503 with "restarting": true (or 502 if container is running but unresponsive)
  2. Proactive: New registry.StartHealthSweep polls Docker API every 15s for all online workspaces → catches dead containers before users notice
  3. Auto-restart: Both liveness monitor and health sweep trigger RestartByID() on offline detection. Per-workspace mutex deduplicates concurrent restart attempts.
• WorkspaceHandler moved from router.Setup to main.go creation so RestartByID is accessible in offline callbacks
• New db.ClearWorkspaceKeys() shared helper replaces 3x duplicated Redis cleanup
• New files: workspace-server/internal/registry/healthsweep.go, healthsweep_test.go (3 tests)
• Files: workspace-server/cmd/server/main.go, workspace-server/internal/handlers/workspace.go, workspace-server/internal/router/router.go, workspace-server/internal/db/redis.go, workspace-server/internal/registry/healthsweep.go

Template Fallback for Missing Templates

• Root cause of auth error: setup-org.sh referenced non-existent org-* templates → containers got empty /configs → fell back to langgraph runtime with anthropic:claude-sonnet-4-6 but no ANTHROPIC_API_KEY
• Fix: Create handler now validates template exists via os.Stat, falls back to {runtime}-default template, then ensureDefaultConfig()
• runtime column added to List/Get API response (scanWorkspaceRow, workspaceListQuery, Get query)
• Files: workspace-server/internal/handlers/workspace.go, workspace-server/internal/handlers/handlers_test.go

Graceful Delegation Error Handling

• Problem: When child workspace fails (auth error, offline), PM forwarded raw error message to user instead of handling gracefully
• Fix (3 layers):
  1. a2a_mcp_server.py: delegate_task detects errors via [A2A_ERROR] sentinel prefix, wraps as DELEGATION FAILED with instructions to try another peer or handle itself
  2. coordinator.py: Strengthened coordination rule 5 — "do NOT forward raw errors to user"
  3. cli_executor.py: Added IMPORTANT block in A2A instructions for delegation failure handling
• Auth errors in CLI executor now retry with exponential backoff (same as rate limits)
• Claude Code adapter: Fixed dict.get("command", "claude") → .get("command") or "claude" for empty string handling
• Files: workspace/a2a_mcp_server.py, workspace/coordinator.py, workspace/cli_executor.py, workspace/adapters/claude_code/adapter.py

Agent Push Messaging (send_message_to_user)

• Feature: Agents can now push messages to the user's canvas chat at any time — not just as A2A responses
• Use case: Agent says "Got it, delegating now...", continues working, then sends results when done
• Platform: New POST /workspaces/:id/notify endpoint → broadcasts AGENT_MESSAGE via WebSocket (BroadcastOnly)
• MCP tool: send_message_to_user in a2a_mcp_server.py — calls notify endpoint
• Canvas: AGENT_MESSAGE handled in global applyEvent → stored in agentMessages map → ChatTab consumes via store subscription (no extra WS connection)
• Prompts: Updated A2A instructions + CLAUDE.md with "RESPOND FAST, FOLLOW UP LATER" rule
• Files: workspace-server/internal/handlers/activity.go, workspace-server/internal/router/router.go, workspace/a2a_mcp_server.py, canvas/src/store/canvas.ts, canvas/src/components/tabs/ChatTab.tsx, workspace/cli_executor.py, workspace-configs-templates/claude-code-default/CLAUDE.md

Remove Default Agent Timeout

• Changed default timeout from 300s to 0 (no timeout) — delegation chains can take arbitrarily long
• Files: workspace-configs-templates/claude-code-default/config.yaml, workspace/config.py, workspace-server/internal/handlers/workspace.go

WebSocket Error Suppression

• Suppressed noisy WebSocket error: {} console.error in socket.ts — onerror fires before onclose and the Event object has no useful info
• Files: canvas/src/store/socket.ts

Setup Script Fix

• Removed dead code copying auth tokens to non-existent org-* template dirs
• Auth token now auto-propagated via claude-code-default template fallback
• Files: setup-org.sh

Remove Default Agent Timeout

• Problem: PM timed out after 300s during delegation chains. Long-running tasks (multi-agent coordination, research) are expected to exceed 5 minutes.
• Fix: Changed default timeout from 300s to 0 (no timeout) in three places:
  • workspace-configs-templates/claude-code-default/config.yaml — template default
  • workspace/config.py — RuntimeConfig.timeout dataclass default + YAML parser default
  • workspace-server/internal/handlers/workspace.go — ensureDefaultConfig generated config
• timeout: 0 → self.config.timeout or None → None → proc.communicate() waits indefinitely
• Files: workspace-configs-templates/claude-code-default/config.yaml, workspace/config.py, workspace-server/internal/handlers/workspace.go

Build Script for Runtime Images

• Problem: Each runtime has its own Dockerfile extending workspace-template:base with pre-installed deps. Manually running docker build for each is error-prone — we shipped with 5-hour-old images and didn't notice.
• Fix: New workspace/build-all.sh — builds base first, then all 6 runtime images in order. Supports selective builds (build-all.sh claude-code langgraph). Handles underscore/hyphen naming mismatch (dir claude_code → tag claude-code). No :latest tag — each runtime uses its own explicit tag.
• Added missing error logging in activity.go List handler (was returning 500 "query failed" without logging the actual SQL error)
• Files: workspace/build-all.sh (new), workspace-server/internal/provisioner/provisioner.go, workspace-server/internal/handlers/activity.go, CLAUDE.md

Codebase Modularization (Major Refactoring)

Split 6 large files (~4,200 lines total) into 22 focused modules. Pure structural — no behavior changes. All tests pass.

Platform handlers:

• workspace.go (978→377 lines) → split out workspace_provision.go (217), workspace_restart.go (173), a2a_proxy.go (251)
• templates.go (814→371 lines) → split out container_files.go (168), template_import.go (175)

Workspace template:

• a2a_mcp_server.py (572→293 lines) → split out a2a_client.py (97), a2a_tools.py (275)

Canvas:

• ConfigTab.tsx (738→310 lines) → split out config/form-inputs.tsx, config/secrets-section.tsx, config/yaml-utils.ts
• ChatTab.tsx (635→340 lines) → split out chat/types.ts, chat/storage.ts, chat/message-parser.ts
• canvas.ts (449→215 lines) → split out canvas-events.ts, canvas-topology.ts, canvas-capabilities.ts

Tier System Simplified (T1/T2/T3, removed T4)

• T1 Sandboxed: No /workspace mount, config only (unchanged)
• T2 Standard: Normal Docker + /workspace mount (unchanged, was identical to T3 before)
• T3 Full Access: --privileged + --pid=host — full machine access for dev team
• T4 removed: EC2 VMs were unimplemented; privileged Docker achieves the same goal
• Updated provisioner switch statement, CreateWorkspaceDialog (3-col grid, no T4), docs/architecture/workspace-tiers.md (full rewrite)
• Files: workspace-server/internal/provisioner/provisioner.go, canvas/src/components/CreateWorkspaceDialog.tsx, docs/architecture/workspace-tiers.md

Config Volume Persistence (Restart no longer overwrites)

• Problem: Restart re-applied claude-code-default template, overwriting user config changes (e.g. model: opus → sonnet)
• Fix: Restart handler skips templates by default. New "apply_template": true flag in restart body for explicit re-application (used when runtime changes).
• RestartByID (auto-restart) also skips templates — passes empty template path
• Files: workspace-server/internal/handlers/workspace_restart.go

Skills Self-Improvement System

• Documented how agents can create persistent skills in /configs/skills/<name>/SKILL.md
• Skills are auto-loaded into system prompt via skills/loader.py
• Skills persist on Docker named volume — survive restarts
• Updated workspace-configs-templates/claude-code-default/CLAUDE.md with skills creation guide
• Trained PM agent to convert operating procedures into skills

Agent Code Fixes (from agent-written code)

• Fixed pytest.ini: removed --cov-fail-under=100 that broke test runner
• Fixed 6 test files: replaced hardcoded /workspace/workspace/ paths with os.path.dirname(__file__) relative paths
• Fixed aes_test.go: test key that wasn't 32 bytes after base64 decode
• Fixed agent_test.go: SQL mock arg count mismatch (2 args for 1-param query)
• Fixed liveness_test.go: unused variable
• Cleaned up .coverage, .coveragerc, __pycache__, index_minimal.ts

Agent Training via A2A

• Sent feedback to PM, Dev Lead, QA Engineer about test-writing rules, path handling, config discipline
• All 3 agents committed rules to persistent memory
• PM + dev team upgraded to Opus 4.6 model, T3 tier
• Marketing/Research teams remain Sonnet, T2

Misc

• .gitignore: Added .claude/worktrees/ to prevent stale worktrees showing as submodule changes

Workspace Pause/Resume (PR #4)

• New POST /workspaces/:id/pause — stops container, sets status='paused', clears Redis keys
• New POST /workspaces/:id/resume — re-provisions from existing config volume
• Health sweep, liveness monitor, and auto-restart all skip paused workspaces
• Canvas: indigo "Paused" status dot, Legend entry, context menu Pause/Resume toggle
• WORKSPACE_PAUSED WebSocket event handled in canvas-events.ts
• Cascade: pausing a parent pauses all descendants (recursive CTE), resuming does the reverse
• Guard: children cannot restart or resume while any ancestor is paused (409 Conflict)
• isParentPaused() recursive helper checks ancestor chain
• Context menu: right-click nested team members now opens correct child menu (not parent's)
• Context menu closes immediately on pause/resume click (before API call, not after)
• Files: workspace-server/internal/handlers/workspace_restart.go, workspace-server/internal/router/router.go, workspace-server/internal/registry/liveness.go, canvas/src/store/canvas-events.ts, canvas/src/components/StatusDot.tsx, canvas/src/components/WorkspaceNode.tsx, canvas/src/components/Legend.tsx, canvas/src/components/ContextMenu.tsx

Files Changed

• canvas/src/components/tabs/ChatTab.tsx
• canvas/src/components/tabs/ConfigTab.tsx
• canvas/src/store/canvas.ts
• canvas/src/store/__tests__/canvas.test.ts
• workspace/a2a_executor.py
• workspace/adapters/langgraph/adapter.py
• workspace/adapters/deepagents/adapter.py
• workspace/adapters/crewai/adapter.py
• workspace/adapters/autogen/adapter.py
• workspace/adapters/openclaw/adapter.py
• workspace/tests/test_a2a_executor.py
• workspace-server/cmd/server/main.go
• workspace-server/internal/db/redis.go
• workspace-server/internal/handlers/workspace.go
• workspace-server/internal/handlers/handlers_test.go
• workspace-server/internal/router/router.go
• workspace-server/internal/registry/healthsweep.go (new)
• workspace-server/internal/registry/healthsweep_test.go (new)
• workspace/a2a_mcp_server.py
• workspace/adapters/claude_code/adapter.py
• workspace/cli_executor.py
• workspace/coordinator.py
• setup-org.sh
• CLAUDE.md
• docs/architecture/provisioner.md
• workspace/config.py
• workspace-configs-templates/claude-code-default/config.yaml
• workspace-configs-templates/claude-code-default/CLAUDE.md
• workspace-server/internal/handlers/activity.go
• canvas/src/store/socket.ts
• docs/architecture/provisioner.md
• workspace-server/internal/provisioner/provisioner.go
• workspace/build-all.sh (new)
• docs/agent-runtime/cli-runtime.md
• docs/agent-runtime/config-format.md
• workspace-server/internal/handlers/workspace_provision.go (new — extracted from workspace.go)
• workspace-server/internal/handlers/workspace_restart.go (new — extracted from workspace.go)
• workspace-server/internal/handlers/a2a_proxy.go (new — extracted from workspace.go)
• workspace-server/internal/handlers/container_files.go (new — extracted from templates.go)
• workspace-server/internal/handlers/template_import.go (new — extracted from templates.go)
• workspace/a2a_client.py (new — extracted from a2a_mcp_server.py)
• workspace/a2a_tools.py (new — extracted from a2a_mcp_server.py)
• workspace/tests/test_mcp_memory.py
• canvas/src/store/canvas-events.ts (new — extracted from canvas.ts)
• canvas/src/store/canvas-topology.ts (new — extracted from canvas.ts)
• canvas/src/store/canvas-capabilities.ts (new — extracted from canvas.ts)
• canvas/src/components/tabs/chat/types.ts (new)
• canvas/src/components/tabs/chat/storage.ts (new)
• canvas/src/components/tabs/chat/message-parser.ts (new)
• canvas/src/components/tabs/chat/index.ts (new)
• canvas/src/components/tabs/config/form-inputs.tsx (new)
• canvas/src/components/tabs/config/secrets-section.tsx (new)
• canvas/src/components/tabs/config/yaml-utils.ts (new)
• canvas/src/components/tabs/config/index.ts (new)