system-planner
Skill0 starsChanged 3 months agoApache-2.0View on GitHub
Own the reusable planning workflow for Phase 2. Use for planning workflow, ExecPlan drafting, slice planning, scope decomposition.
In this repository78 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .agents/skills/go-skills-knowledge-pack/AGENTS.mdAGENTS.md
- .agents/skills/rust-skills-knowledge-pack/AGENTS.mdAGENTS.md
- .agents/skills/typescript-skills-knowledge-pack/AGENTS.mdAGENTS.md
- .claude/skills/go-skills-knowledge-pack/AGENTS.mdAGENTS.md
- .claude/skills/rust-skills-knowledge-pack/AGENTS.mdAGENTS.md
- .claude/skills/typescript-skills-knowledge-pack/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .cursor/rules/00-general.mdcCursor rule
- .cursor/rules/revharness-critical.mdcCursor rule
- .cursor/rules/revharness-detailed.mdcCursor rule
- auto-orchestratorSkill
- baseline-protectionSkill
- client-distribution-readinessSkill
- cloudflare-deploy-guardSkill
- codex-app-server-guardSkill
- codex-app-server-product-integrationSkill
- codex-callerSkill
- context-compactorSkill
- cursor-callerSkill
- delegated-development-disciplineSkill
- deploy-command-disciplineSkill
- design-principlesSkill
- development-constitutionSkill
- development-junk-cleanupSkill
- go-skills-knowledge-packSkill
- harness-official-docs-updateSkill
- naming-normalization-guardSkill
- orchestrator-bootstrapSkill
- payload-cms-deploy-guardSkill
- production-function-implementerSkill
- research-handoffSkill
- rev-harness-lifecycleSkill
- review-workflowSkill
- root-cause-first-disciplineSkill
- rust-skills-knowledge-packSkill
- scope-guardSkill
- self-growth-proposal-triageSkill
- shared-client-wrapper-safetySkill
- slice-designerSkill
- staff-code-reviewerSkill
- structured-mentorSkill
- supabase-deploy-guardSkill
- typescript-skills-knowledge-packSkill
- auto-orchestratorSkill
- baseline-protectionSkill
- client-distribution-readinessSkill
- cloudflare-deploy-guardSkill
- codex-app-server-guardSkill
- codex-app-server-product-integrationSkill
- codex-callerSkill
- context-compactorSkill
- cursor-callerSkill
- delegated-development-disciplineSkill
- deploy-command-disciplineSkill
- design-principlesSkill
- development-constitutionSkill
- development-junk-cleanupSkill
- go-skills-knowledge-packSkill
- harness-official-docs-updateSkill
- The first 60 shown.
The file
Files
Skill: System Planner
Phase 2 の planning と plan handoff はこの skill が所有する。要件整理、スコープ分解、検証方針、次の担当への受け渡しを定義する。
使う場面
- ExecPlan が未作成
- 既存 plan が古い、またはスコープ変更で更新が必要
- 調査結果を実装計画へ落とし込みたい
期待成果物
.agent/active/plan_YYYYMMDD_HHMM_<task>.md- 実装/レビューに渡す要約プロンプト
- 依存関係、リスク、検証項目の整理
- required checks の exact command / artifact landing path / completion boundary
Phase 2 の境界
.codex/agents/system_planner.tomlは native subagent preset。caller-facing runtime contract ではない。- 初回設計 / ExecPlan drafting / ExecPlan review planning は
.agent/registry/model_policy.jsonのinitial_execplan_designlane を正本とし、gpt-5.5+xhigh+cachedで扱う。通常の docs-only / light planning、typo、bookkeeping と混同してmediumへ落とさない。 - 人や script から Codex を呼ぶ入口は、引き続き
scripts/codex-wrapper.sh --role <standard|research|coder|high-coder|reviewer>を使う。 - Codex 呼び出し構文そのものは
codex-callerskill を参照する。
Planning Contract
- planning 前に
docs/manual/verification-truth-matrix.mdを読み、slice record を作る。 - 各 slice には
in-scope/out-of-scope/required checks/completion boundaryを必ず記録する。 - stable truth は manual / README / skill に置き、session 依存の内容は
.agent/active/plan_*.md/.agent/active/sow/*.md/ handoff prompt /.claude/tmp/**に置く。 - stable document に dated rerun 結果、個別 run artifact、暫定 blocker を混在させない。
- Codex / Claude Code / prompting / Goal / subagent / skill / hook / settings behavior を変える plan では、
.claude/skills/harness-official-docs-update/SKILL.mdまたはresearch-handoffの official-docs provenance を先に記録する。 - upstream recommendation は local authority mapping に変換してから slice 化する。公式 docs だけを根拠に wrapper / acceptance / security boundary を上書きしない。
Deterministic Verification Commands
planner は change surface ごとに、最低でも次の exact commands を plan に書く。
| Change surface | Required commands |
| 純粋な説明文書のみ | git diff --check -- <files> |
| コマンド、パス、gate、review 条件、運用例を追加・変更する文書 | git diff --check -- <files> と、追加・変更した参照先ごとの test -e <path> |
| script / wrapper / 実行入口 | git diff --check -- <files>、bash -n <file>、変更した入口の非破壊 help / syntax / existence probe |
- planner は
<files>と<path>を省略せず、実際の対象ファイルに展開したコマンドを残す。 - command / script 契約に触れる slice では、
bash <script> --helpのような非破壊 probe も required checks に含める。 - release gate が必要な surface は、補助 gate ではなく
bash test/integration/harness_release_gate.shを authoritative gate として記録する。
Artifact Landing Paths
- slice 定義、required checks、completion boundary:
.agent/active/plan_YYYYMMDD_HHMM_<task>.md - 実行中の blocker、rerun 判断、補足 evidence:
.agent/active/sow/*.md - reviewer / coder への handoff 文面:
.agent/active/prompts/*.md - command stdout / stderr などの run artifact:
.claude/tmp/**
Checkpoint Blockers
- broad task が未分解で、scope-bounded slice を定義できない
- required checks の exact command が書けない
- artifact landing path が決まっていない
- stable と volatile の置き場が混在している
- 環境制約で required checks を実行できないのに reroute / block 方針が未記録
上記のいずれかがある場合、planner は completion boundary を開けず BLOCK として返す。
運用メモ
- Claude Orchestrator の effort 既定値は
medium。 - 軽量な plan 整形なら Codex は
--role standardを検討できる。ただし初回 ExecPlan 設計、mutation-authorizing ExecPlan drafting、ExecPlan review planning は nativesystem_planner/plan_reviewerの xhigh lane を使う。 - 外部の不確実性が残る場合は
research-handoffを先に通す。
参照
docs/manual/verification-truth-matrix.mddocs/roles/orchestrator.md.claude/skills/auto-orchestrator/SKILL.md.claude/skills/codex-caller/SKILL.md.claude/skills/harness-official-docs-update/SKILL.md
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

