mirror of
https://github.com/merlinhu1/codex-game-studio.git
synced 2026-08-25 07:54:34 +02:00
7e8846a7fb
* feat: add deterministic model tier routing Define Sol, Terra, and Luna policies with fixed fallbacks, explicit surface assignments, runtime precedence, and Luna escalation. Validate generated metadata and expose tier overrides across role, task, and orchestration commands. * fix: infer model tiers from surface models * fix: rebalance prompt model routing * fix: decouple prompt effort routing --------- Co-authored-by: MerlinH <merlinh221@gmail.com>
2.5 KiB
2.5 KiB
name, description, model, model_reasoning_effort, argument-hint, primary-agent, tool-policy, isolation, source-reference, source-hash, user-invocable
| name | description | model | model_reasoning_effort | argument-hint | primary-agent | tool-policy | isolation | source-reference | source-hash | user-invocable |
|---|---|---|---|---|---|---|---|---|---|---|
| cgs-retrospective | Use for retrospective tasks that facilitate a retrospective with what worked, what failed, root causes, actions, and owner follow-up; produce verification evidence, changed or proposed files, and handoff boundaries. | gpt-5.6-luna | low | Describe the retrospective objective, target files/assets, constraints, and verification evidence. | producer | read/edit/shell/tests/git as needed within the repository write policy | repository-root; no init-time generated prompt bodies | .claude/skills/retrospective/SKILL.md | 60f9525887c94e4cf8239de3a2078389e6d6b19dd9f10705772490650c11ffda | true |
Codex Game Studio Retrospective
Use this skill for retrospective work in Template Game.
Objective
Facilitate a retrospective with what worked, what failed, root causes, actions, and owner follow-up.
Inputs
- AGENTS.md
- .codex/studio.json
- task-relevant files named by the user or task record
- production/timeline.md
- docs/market-overview.md
- tests/
Arguments
- Objective or user request.
- Target files, scenes, assets, or docs.
- Constraints, deadlines, acceptance criteria, and verification command when known.
Procedure
- Clarify the requested facilitate a retrospective with what worked, what failed, root causes, actions, and owner follow-up. and identify the current project stage.
- Collect evidence for Went Well, Did Not Work, Root Cause.
- Separate blockers from warnings and include rollback or deferral options.
- Produce the requested artifact or review with clear file paths and verification evidence.
Write Targets
- production/
- docs/
Output Contract
- Summary
- Went Well
- Did Not Work
- Root Cause
- Action Item
- Blocking issues
- Changed files or proposed files
- Verification evidence
- Next owner or decision
Quality Gates
- Went Well
- Did Not Work
- Root Cause
- Action Item
- Scope remains bounded to the current task and project stage.
- Report labels unverified assumptions separately from evidence.
Decision Gates
- Continue only when the expected output can be verified or clearly labeled as a plan.
- Escalate to producer or qa-playtester when scope, ownership, or acceptance evidence is ambiguous.
- Stop before broad rewrites, generated prompt mirrors, or hidden lifecycle behavior.
Handoff
Report changed files, verification evidence, remaining risks, and the next owner or decision. Do not imply hidden hooks or autonomous background work.