fix(leads): put dev-team leads on Claude Opus, not MiniMax #10

hongming-pc2 · 2026-05-10T09:28:09Z

hongming-pc2 commented

2026-05-10 09:28:09 +00:00

Per the model-tiering directive (sub-team leads → Claude Opus subscription, ICs → MiniMax-M2.7), the six dev-team leads run model: opus instead of MiniMax-M2.7.

Why

PR #9 ("wire dev-department LLM env to MiniMax-M2.7") set the entire dev-tree to MiniMax, including the leads. The leads should be on Opus — opus is the anthropic-oauth entry in config.yamls runtime_config.models (auth via CLAUDE_CODE_OAUTH_TOKEN, no base_url).

Changed

dev-lead, app-lead, core-lead, cp-lead, infra-lead, sdk-lead → model: opus.
Unchanged: integration-tester, release-manager (ICs, stay on MiniMax); fullstack-engineer + triage-operator were already model: opus.

Runtime caveat (so reviewers know the full picture)

molecule_runtime/config.py reads the MODEL_PROVIDER env var as the picked model id (misnomer), at higher precedence than this model: field. The live fix on the PC2 platform is in per-workspace workspace_secrets (MODEL_PROVIDER=opus, CLAUDE_CODE_OAUTH_TOKEN=<oat>, blanked ANTHROPIC_AUTH_TOKEN/ANTHROPIC_BASE_URL so the MiniMax-flavored global_secrets dont leak in) — verified: all 6 leads now produce "model":"claude-opus-4-7" transcript output with no errors. This PR makes the template self-consistent so a fresh org/import produces the right thing.

The operator-host lead persona env files (/etc/molecule-bootstrap/personas/<lead>/env) still carry the stale MODEL_PROVIDER=claude-code (the bug — it made the claude CLI 404 on --model claude-code) and need the same correction out of band; they arent tracked in any repo.

🤖 Generated with Claude Code

Per the model-tiering directive (sub-team **leads → Claude Opus subscription**, ICs → MiniMax-M2.7), the six dev-team leads run `model: opus` instead of `MiniMax-M2.7`. ### Why PR #9 ("wire dev-department LLM env to MiniMax-M2.7") set the *entire* dev-tree to MiniMax, including the leads. The leads should be on Opus — `opus` is the `anthropic-oauth` entry in `config.yaml`s `runtime_config.models` (auth via `CLAUDE_CODE_OAUTH_TOKEN`, no `base_url`). ### Changed `dev-lead`, `app-lead`, `core-lead`, `cp-lead`, `infra-lead`, `sdk-lead` → `model: opus`. Unchanged: `integration-tester`, `release-manager` (ICs, stay on MiniMax); `fullstack-engineer` + `triage-operator` were already `model: opus`. ### Runtime caveat (so reviewers know the full picture) `molecule_runtime/config.py` reads the **`MODEL_PROVIDER` env var as the *picked model id*** (misnomer), at higher precedence than this `model:` field. The live fix on the PC2 platform is in per-workspace `workspace_secrets` (`MODEL_PROVIDER=opus`, `CLAUDE_CODE_OAUTH_TOKEN=<oat>`, blanked `ANTHROPIC_AUTH_TOKEN`/`ANTHROPIC_BASE_URL` so the MiniMax-flavored `global_secrets` dont leak in) — verified: all 6 leads now produce `"model":"claude-opus-4-7"` transcript output with no errors. This PR makes the *template* self-consistent so a fresh `org/import` produces the right thing. The operator-host lead persona env files (`/etc/molecule-bootstrap/personas/<lead>/env`) still carry the stale `MODEL_PROVIDER=claude-code` (the bug — it made the `claude` CLI 404 on `--model claude-code`) and need the same correction out of band; they arent tracked in any repo. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

hongming-pc2 added 1 commit 2026-05-10 09:28:12 +00:00

fix(leads): put dev-team leads on Claude Opus, not MiniMax aa50287f8b

PR #9 wired the entire dev-department to MiniMax-M2.7, including the six
sub-team leads. Per the model-tiering directive (leads → Claude Opus
subscription, ICs → MiniMax-M2.7), the leads must run `model: opus`
(the anthropic-oauth entry in config.yaml's runtime_config.models —
auth via CLAUDE_CODE_OAUTH_TOKEN, no base_url).

Changed: dev-lead, app-lead, core-lead, cp-lead, infra-lead, sdk-lead.
Unchanged: integration-tester, release-manager (ICs, stay on MiniMax);
fullstack-engineer + triage-operator already `model: opus`.

NB: molecule_runtime reads the MODEL_PROVIDER env var as the *picked
model id* (misnomer), at higher precedence than this `model:` field.
The deployed fix on the PC2 platform also sets per-workspace
workspace_secrets MODEL_PROVIDER=opus + CLAUDE_CODE_OAUTH_TOKEN and
blanks ANTHROPIC_AUTH_TOKEN/ANTHROPIC_BASE_URL (so the MiniMax-flavored
global_secrets don't leak in). The operator-host lead persona env files
(/etc/molecule-bootstrap/personas/<lead>/env) still carry the stale
MODEL_PROVIDER=claude-code (which made the claude CLI 404 on
`--model claude-code`) and need the same correction out of band — they
aren't tracked in any repo.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

claude-ceo-assistant added the

tier:low

label 2026-05-10 10:18:18 +00:00

claude-ceo-assistant merged commit 3992268132 into main

2026-05-10 10:18:19 +00:00

claude-ceo-assistant referenced this issue from a commit

2026-05-10 10:18:20 +00:00

Merge pull request 'fix(leads): put dev-team leads on Claude Opus, not MiniMax' (#10) from fix/leads-on-opus into main

claude-ceo-assistant referenced this issue from a commit

2026-05-10 11:28:53 +00:00

feat(defaults): add molecule-skill-five-axis-review plugin (#6)

Sign in to join this conversation.