mochan-tk

Designs supported Codex Scheduled tasks, repository hooks, MCP connections, and optional GitHub Action integrations with least privilege and an explicit human trust boundary. Use when a recurring task, event-driven check, external tool connection, scheduled…

Collects the minimum project context needed for later distillation while preserving source provenance and governed originals. Use when new documents, interviews, meetings, external-tool records, research, or contradictory source material must enter the…

Distills provenance-linked context into reviewable requirements, decisions, non-goals, and vocabulary with an explicit agreement-merge boundary. Use after collection, before planning from new project knowledge, when sources conflict, or when durable guidance…

Builds and maintains the executable GitHub Issue graph of Epics, Task sub-issues, dependencies, origin links, and the actionable frontier. Use when creating or splitting work, wiring blockers, planning a rolling wave, deciding what can run next or in…

Guides Codex through measured onboarding of a new or existing repository, including command characterization, legacy characterization-test work, post-agreement project tuning, and license evidence. Use when a template is first adopted, repository commands or…

Turns repeated project or agent-system friction into the smallest traceable improvement to guidance, skills, templates, tests, or gates while controlling always-on context and upstream flow. Use on a first occurrence that needs a retro candidate, a second…

Runs, delegates, coordinates, and crash-resumes GitHub Tasks and sibling Epic phases through Codex subagents while keeping ownership, plans, evidence, and outcomes authoritative on GitHub. Use when starting or delegating a Task, conducting a program or Epic,…

Routes a GitHub Task to the appropriate Codex app, CLI, IDE, or cloud surface and records an advisory role and reasoning tier without pinning a model. Use while drafting a Task, assigning its single exec label, changing environments after new evidence, or…

Tune this generic scaffold to a concrete target project — inventory the repo, interview only the gaps, verify every command by actually running it, fill or remove every CUSTOMIZE block, and deliver one evidence-backed onboarding PR. Use this whenever the…

Protocol for running work through parent/child agent sessions (e.g., the GitHub Copilot app's session tree, or any orchestrator dispatching cloud-agent runs). Use this whenever a session spawns or reports to another session, when starting work on a Task issue…

Decide where and with what each Task issue should run — execution surface (exec:cloud / exec:app / exec:cli / exec:ide label), suggested agent role, and suggested model/reasoning effort. Use this while decomposing an Epic, when filling the Routing block of…

Build and maintain the executable plan as a GitHub issue graph — Epics, just-in-time Task decomposition, blocked-by dependencies, the actionable frontier, and replanning after deviations. Use this whenever creating or splitting issues, deciding what can run…

Convert repeated failures, recurring review comments, and incidents into permanent system improvements — targeted diffs to instructions, skills, agreements, templates, or CI, delivered as small `retro:` PRs and logged. Use this when the same class of mistake…

Prove that work meets its Task issue — layered verification gates, the pre-PR checklist, the acceptance-criteria evidence table, and CI-failure triage. Use this before marking any PR ready for review, when writing the Verification section of a task brief,…

Land raw project information — meeting notes, existing specs, domain research, external tool exports, interview answers — into .github/docs/context/ with provenance, so later phases and stateless agents can use it. Use this at project kickoff, whenever new…

Turn raw material in .github/docs/context/ into reviewed agreements (requirements with REQ-### IDs, ADRs, glossary, non-goals) and place each piece of knowledge in the right context tier — always-on instructions, path-scoped instructions, or on-demand…

Assess and propose repository onboarding safely. Use when adopting or re-tuning this kit. Do not use for ordinary feature implementation.

Collect bounded source context with provenance. Use when requirements must be gathered. Do not use to approve agreements or execute implementation.

Distill pinned context into reviewable agreements. Use when evidence must become requirements or decisions. Do not use for raw collection.

Maintain the canonical Issue graph and rolling plan. Use when decomposing or replanning durable work. Do not use to execute a Task.

Convert failures into stronger controls. Use after recurrence, a material incident or rejected PR, or scheduled hygiene. Do not use for one-off repair or feedback transport.

Coordinate durable Task attempts and recovery. Use when supervising bounded execution contexts. Do not use as a runtime identity guarantee.

Recommend a bounded execution route from evidence. Use when a planned Task needs placement. Do not use to claim runtime support or dispatch.

Verify acceptance against exact current evidence. Use when defining Verification sections, triaging CI, or judging a Task or pull request. Do not use to approve agreements or final completion.

Turn raw material in docs/context/ into reviewed agreements (requirements with REQ-### IDs, ADRs, glossary, non-goals) and place each piece of knowledge in the right context tier — always-on instructions, path-scoped instructions, or on-demand skills/docs.…

Tune this generic scaffold to a concrete target project — inventory the repo, interview only the gaps, verify every command by actually running it, fill or remove every CUSTOMIZE block, and deliver one evidence-backed onboarding PR. Use this whenever the…

Protocol for running work through parent/child agent sessions (e.g., a Claude Code orchestrator session spawning worktree-isolated child sessions or subagents, or any orchestrator dispatching cloud-agent runs). Use this whenever a session spawns or reports to…

Prove that work meets its Task issue — layered verification gates, the pre-PR checklist, the acceptance-criteria evidence table, and CI-failure triage. Use this before marking any PR ready for review, when writing the Verification section of a task brief,…

Build and maintain the executable plan as a GitHub issue graph — Epics, just-in-time Task decomposition, blocked-by dependencies, the actionable frontier, and replanning after deviations. Use this whenever creating or splitting issues, deciding what can run…

Land raw project information — meeting notes, existing specs, domain research, external tool exports, interview answers — into docs/context/ with provenance, so later phases and stateless agents can use it. Use this at project kickoff, whenever new source…

Convert repeated failures, recurring review comments, and incidents into permanent system improvements — targeted diffs to instructions, skills, agreements, templates, or CI, delivered as small `retro:` PRs and logged. Use this when the same class of mistake…

Decide where and with what each Task issue should run — execution surface (exec:cloud / exec:app / exec:cli / exec:ide label), suggested agent role, and suggested model/reasoning effort. Use this while decomposing an Epic, when filling the Routing block of…

Land raw project information — meeting notes, existing specs, domain research, external tool exports, interview answers — into docs/context/ with provenance, so later phases and stateless agents can use it. Use this at project kickoff, whenever new source…

Turn raw material in docs/context/ into reviewed agreements (requirements with REQ-### IDs, ADRs, glossary, non-goals) and place each piece of knowledge in the right context tier — always-on instructions, path-scoped instructions, or on-demand skills/docs.…

Build and maintain the executable plan as a GitHub issue graph — Epics, just-in-time Task decomposition, blocked-by dependencies, the actionable frontier, and replanning after deviations. Use this whenever creating or splitting issues, deciding what can run…

Tune this generic scaffold to a concrete target project — inventory the repo, interview only the gaps, verify every command by actually running it, fill or remove every CUSTOMIZE block, and deliver one evidence-backed onboarding PR. Use this whenever the…

Convert repeated failures, recurring review comments, and incidents into permanent system improvements — targeted diffs to instructions, skills, agreements, templates, or CI, delivered as small `retro:` PRs and logged. Use this when the same class of mistake…

Protocol for running work through parent/child agent sessions (e.g., the GitHub Copilot app's session tree, or any orchestrator dispatching cloud-agent runs). Use this whenever a session spawns or reports to another session, when starting work on a Task issue…

Decide where and with what each Task issue should run — execution surface (exec:cloud / exec:app / exec:cli / exec:ide label), suggested agent role, and suggested model/reasoning effort. Use this while decomposing an Epic, when filling the Routing block of…

Prove that work meets its Task issue — layered verification gates, the pre-PR checklist, the acceptance-criteria evidence table, and CI-failure triage. Use this before marking any PR ready for review, when writing the Verification section of a task brief,…

Tune this generic scaffold to a concrete target project — inventory the repo, interview only the gaps, verify every command by actually running it, fill or remove every CUSTOMIZE block, and deliver one evidence-backed onboarding PR. Use this whenever the…

Build and maintain the executable plan as a GitHub issue graph — Epics, just-in-time Task decomposition, blocked-by dependencies, the actionable frontier, and replanning after deviations. Use this whenever creating or splitting issues, deciding what can run…

Protocol for running work through parent/child agent sessions (e.g., the GitHub Copilot app's session tree, or any orchestrator dispatching cloud-agent runs). Use this whenever a session spawns or reports to another session, when starting work on a Task issue…

Decide where and with what each Task issue should run — execution surface (exec:cloud / exec:app / exec:cli / exec:ide label), suggested agent role, and suggested model/reasoning effort. Use this while decomposing an Epic, when filling the Routing block of…

Convert repeated failures, recurring review comments, and incidents into permanent system improvements — targeted diffs to instructions, skills, agreements, templates, or CI, delivered as small `retro:` PRs and logged. Use this when the same class of mistake…

Prove that work meets its Task issue — layered verification gates, the pre-PR checklist, the acceptance-criteria evidence table, and CI-failure triage. Use this before marking any PR ready for review, when writing the Verification section of a task brief,…

Use when: implementing frontend web apps with HTML, CSS, TypeScript, React, accessibility, responsive layout, or performance concerns.

Use when: reviewing or fixing frontend UI layout, responsive design, accessibility, visual consistency, or mobile display issues.

Use when: testing frontend behavior in a browser with Playwright, screenshots, console logs, forms, navigation, or responsive checks.

Build and maintain the executable plan as a GitHub issue graph — Epics, just-in-time Task decomposition, blocked-by dependencies, the actionable frontier, and replanning after deviations. Use this whenever creating or splitting issues, deciding what can run…

Protocol for running work through parent/child agent sessions (e.g., the GitHub Copilot app's session tree, or any orchestrator dispatching cloud-agent runs). Use this whenever a session spawns or reports to another session, when starting work on a Task issue…

書籍原稿(manuscript/配下のMarkdown)のレビューを行う。原稿のレビュー・推敲・校正・読みやすさ確認を依頼されたとき、またはPRで原稿の変更を確認するときに使う。出版社ルール(context/03_style-guide.md)準拠の観点で、文章・コード・図表・構成・AI臭を確認し、具体的な改善案つきで指摘する。