pma

Project development lifecycle management with a strict three-phase workflow (investigate -> proposal -> implement), file-based plan tracking in docs/plan/, task tracking in docs/task/, and claim-before-work multi-agent coordination. Use when handling feature development, bug fixes, refactors, planning, progress tracking, or multi-agent execution in an existing codebase. Supports English and Chinese project templates.

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "pma" with this command: npx skills add zzci/skills/zzci-skills-pma

PMA - Project Management Assistant

Run delivery work with clear gates, minimal diffs, and explicit task/plan tracking.

Hard Rules

  1. Use one project content language for docs, task files, and commit messages (follow user preference, otherwise follow existing project language).
  2. Read before write: inspect call chains, related config/tests, and recent changelog context before editing logic.
  3. Make only the minimal requested changes; do not add unrequested refactors or features.
  4. Never use plan mode (EnterPlanMode, mode: "plan"). Manage plans in docs/plan/ files only.
  5. Do not implement before explicit confirmation (proceed).
  6. Use English filenames only, even when project content is Chinese (for example: architecture.md, changelog.md).

Three-Phase Workflow

Phase 1: Investigation

  1. Trace upstream/downstream call chains, symbol references, and types.
  2. Search related code, config, tests, migrations, and docs.
  3. Read the tail of docs/changelog.md for recent context.
  4. Find or create the matching task in docs/task/index.md and claim it ([-]).

Non-trivial task rule:

  • If the change touches >=3 files or crosses modules, create docs/plan/PLAN-NNN.md and write findings to the context section.

Phase 2: Proposal

Output these items, then stop for approval:

  • Current state
  • Proposal
  • Risks
  • Scope
  • Alternatives (if multiple approaches exist)

For non-trivial tasks:

  • Complete remaining sections in PLAN-NNN.md.
  • Append one line to docs/plan/index.md with [ ].
  • Wait for user annotations and address all of them before implementation.

Phase 3: Implement -> Verify -> Record

Only after approval:

  1. If a plan exists, set plan index marker to [-] and detail status to implementing.
  2. Implement step by step according to the approved proposal.
  3. Run focused self-verification.
  4. Set task index marker to [x] and task detail status to completed.
  5. If a plan exists, set plan index marker to [x] and plan detail status to completed.
  6. Update changelog as needed.

Task and Plan Files

Use these canonical references instead of redefining formats in-place:

Required structure:

  • docs/task/index.md: one-line task entries
  • docs/task/PREFIX-NNN.md: task detail files
  • docs/plan/index.md: one-line plan entries
  • docs/plan/PLAN-NNN.md: plan detail files

Claim-Before-Work (Multi-Agent Safety)

Before writing any implementation code:

  1. Read docs/task/index.md; for [-] items, read detail owner.
  2. If another agent owns the in-progress task, skip it.
  3. Claim atomically:
    • Update task index [ ] -> [-]
    • Update task detail status -> in_progress, set owner
    • Call TaskUpdate(status: "in_progress", owner: "<agent>") if task tools are available
  4. Start implementation only after the claim is fully written.

On completion:

  • Set task index [-] -> [x]
  • Set task detail status -> completed
  • Call TaskUpdate(status: "completed") if task tools are available

On close/won't do:

  • Set task index to [~]
  • Set task detail status -> closed and record reason
  • Call TaskUpdate(status: "deleted") if task tools are available

Sync Rules

Task status updates are immediate, never deferred.

  • Primary source is files in docs/task/ and docs/plan/.
  • If TaskCreate/TaskUpdate tools are available, keep tool state in sync with file state.
  • If task tools are unavailable, continue with file-only sync and state this in the progress update.

Session checklist:

  1. Session start: read docs/task/index.md, active task details, and docs/plan/index.md.
  2. New task: create detail file first, then append index line.
  3. Before work: complete Claim-Before-Work.
  4. Session end: verify statuses are written and update index header date.

Documentation System

Canonical structure:

docs/
├── task/
│   ├── index.md
│   └── PREFIX-NNN.md
├── plan/
│   ├── index.md
│   └── PLAN-NNN.md
├── architecture.md
└── changelog.md
  • Use English sections for English-language projects.
  • Use Chinese template sections for Chinese-language projects while keeping filenames in English.
  • Write investigation findings into the plan context section.
  • Do not create extra report files; temporary files go to ./tmp/.

Changelog Conventions

Entry format:

## YYYY-MM-DD HH:MM [tag]

[content]

Recommended tags:

  • [progress], [BUG-P0], [BUG-P1], [pitfall], [decision]

Project Initialization

On first use in a project:

  1. Ensure CLAUDE.md contains a ## Project Development section that references /pma and the three-phase workflow.
  2. Ensure AGENTS.md contains the same section.
  3. Ensure docs/task/index.md exists (initialize from docs/task-format.md).
  4. Ensure docs/plan/index.md exists (initialize from docs/plan-format.md).
  5. Ensure core docs exist (architecture.md + changelog.md), and keep content language aligned with the project.

Tools and Security

  • Prefer semantic code navigation tools for architecture understanding.
  • Check official docs (e.g., Context7) before using third-party APIs.
  • Use code/web search tools for examples and troubleshooting when needed.
  • Verify UI behavior after UI edits.
  • Keep secrets in .env; never hardcode or log secrets.
  • Mark risky commands with an explicit warning comment.

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

General

ptask

No summary provided by upstream source.

Repository SourceNeeds Review
Coding

Self Updater

⭐ OPEN SOURCE! GitHub: github.com/GhostDragon124/openclaw-self-updater ⭐ ONLY skill with Cron-aware + Idle detection! Auto-updates OpenClaw core & skills, an...

Registry SourceRecently Updated
1221Profile unavailable
Coding

ClawHub CLI Assistant

Use the ClawHub CLI to publish, inspect, version, update, sync, and troubleshoot OpenClaw skills from the terminal.

Registry SourceRecently Updated
1.9K2Profile unavailable
Coding

SkillTree Learning Progress Tracker

Track learning across topics like an RPG skill tree. Prerequisites, milestones, suggested next steps. Gamified learning path.

Registry SourceRecently Updated
900Profile unavailable