Update org harness/model/effort and PR review policy #37

Merged
privilegedescalation-ceo[bot] merged 1 commits from org-adjustments-harness-model-policy into main 2026-03-25 00:01:21 +00:00
privilegedescalation-ceo[bot] commented 2026-03-24 18:52:58 +00:00 (Migrated from github.com)

Summary

Board directive (PRI-827) to update agent harness, model, effort, and review policy:

Harness/Model/Effort changes:

  • CEO (Countess): No change — already Claude Opus 4.6 medium effort
  • CTO (Nancy): Effort medium → high
  • QA (Regina): opencode_local MiniMax M2.7 → claude_local Claude Sonnet 4.6 high effort
  • Engineer (Gandalf): claude_local Sonnet 4.6 → opencode_local MiniMax M2.7
  • DevOps (Hugh): claude_local Sonnet 4.6 → opencode_local MiniMax M2.7

Policy changes:

  • All work is broken down and distributed by the CTO for engineers to execute
  • PR review order is now mandatory: QA reviews first, CTO reviews second
  • CTO must not review/approve a PR before QA has approved
  • QA should refuse to review if CTO doesn't follow this process

Files changed:

  • cto/CONFIG.md — effort to high
  • engineering/regina/CONFIG.md — adapter switch to claude_local, Sonnet 4.6, high effort
  • engineering/gandalf/CONFIG.md — adapter switch to opencode_local, MiniMax M2.7
  • engineering/hugh/CONFIG.md — adapter switch to opencode_local, MiniMax M2.7
  • cto/SOUL.md — QA-first review policy, work distribution rule
  • engineering/regina/SOUL.md — QA-first enforcement rule
  • POLICIES.md — updated PR workflow and work distribution policy
  • COMPANY.md — updated agent roster table and restoration docs
  • CLAUDE.md — updated adapter references and operational notes

Test plan

  • Verify CONFIG.md adapter configs are valid JSON
  • Verify POLICIES.md review workflow is clear and unambiguous
  • Verify COMPANY.md roster table matches CONFIG.md files
  • Verify no role/intent changes — only harness/model/effort and policy

🤖 Generated with Claude Code

cc @cpfarhood

## Summary Board directive (PRI-827) to update agent harness, model, effort, and review policy: **Harness/Model/Effort changes:** - **CEO (Countess)**: No change — already Claude Opus 4.6 medium effort - **CTO (Nancy)**: Effort medium → **high** - **QA (Regina)**: opencode_local MiniMax M2.7 → **claude_local Claude Sonnet 4.6 high effort** - **Engineer (Gandalf)**: claude_local Sonnet 4.6 → **opencode_local MiniMax M2.7** - **DevOps (Hugh)**: claude_local Sonnet 4.6 → **opencode_local MiniMax M2.7** **Policy changes:** - All work is broken down and distributed by the CTO for engineers to execute - PR review order is now mandatory: **QA reviews first, CTO reviews second** - CTO must not review/approve a PR before QA has approved - QA should refuse to review if CTO doesn't follow this process **Files changed:** - `cto/CONFIG.md` — effort to high - `engineering/regina/CONFIG.md` — adapter switch to claude_local, Sonnet 4.6, high effort - `engineering/gandalf/CONFIG.md` — adapter switch to opencode_local, MiniMax M2.7 - `engineering/hugh/CONFIG.md` — adapter switch to opencode_local, MiniMax M2.7 - `cto/SOUL.md` — QA-first review policy, work distribution rule - `engineering/regina/SOUL.md` — QA-first enforcement rule - `POLICIES.md` — updated PR workflow and work distribution policy - `COMPANY.md` — updated agent roster table and restoration docs - `CLAUDE.md` — updated adapter references and operational notes ## Test plan - [ ] Verify CONFIG.md adapter configs are valid JSON - [ ] Verify POLICIES.md review workflow is clear and unambiguous - [ ] Verify COMPANY.md roster table matches CONFIG.md files - [ ] Verify no role/intent changes — only harness/model/effort and policy 🤖 Generated with [Claude Code](https://claude.com/claude-code) cc @cpfarhood
Sign in to join this conversation.