project-ops

name: project-ops description: Unified long-term project operations for OpenClaw. Use when user asks to list active projects, open a numbered project menu, choose/continue a project by id or index, update status/progress, assign agents quickly, kickoff a new push cycle with a standard log template, generate daily push suggestions from repo git status (blocker-priority first), archive half-finished work in a consistent format, or enforce standardized project documentation under .projects/{id}.

Project Ops v1.4

Single source of truth: workspace/.projects/

Preferred Command Interface

Use wrapper: skills/project-ops/scripts/project.ps1

# /project -> menu
powershell -ExecutionPolicy Bypass -File "skills/project-ops/scripts/project.ps1"

# /project N -> choose by index
powershell -ExecutionPolicy Bypass -File "skills/project-ops/scripts/project.ps1" 2

# /project suggest (today push suggestions by blocker priority)
powershell -ExecutionPolicy Bypass -File "skills/project-ops/scripts/project.ps1" suggest

# alias
powershell -ExecutionPolicy Bypass -File "skills/project-ops/scripts/project.ps1" today

# /project push <project-id>
powershell -ExecutionPolicy Bypass -File "skills/project-ops/scripts/project.ps1" push openclaw-dashboard

# /project log <project-id> <note>
powershell -ExecutionPolicy Bypass -File "skills/project-ops/scripts/project.ps1" log openclaw-dashboard "完成接口核对，记录阶段快照"

# /project archive <project-id> [note]
powershell -ExecutionPolicy Bypass -File "skills/project-ops/scripts/project.ps1" archive openclaw-dashboard "暂缓到下个迭代"

Suggestion Engine (v1.4)

/project suggest reads each project's repoPath and runs:

• git status --porcelain -b

Ranking factors:

1. Project status base (blocked > active > on_hold)
2. Merge conflicts (unmerged) heavy boost
3. Remote behind (behind) medium boost
4. Dirty tree (staged/modified) medium boost
5. Missing nextAction management boost

Output includes rank, reason signals, git summary, immediate action.

Documentation Standard (XiaoZ-derived)

Use this standard for every managed project:

• references/project-doc-standards.md

Canonical docs under .projects/<id>/:

• project.json (machine-readable truth)
• MASTER_PLAN.md (long-range strategy + milestones)
• NEXT_SESSION_PLAN.md (current session target + gate)
• Sxx/IMPLEMENTATION_REPORT.md (session acceptance report)
• logs/YYYY-MM-DD.md (append-only daily trace)

Templates:

• assets/MASTER_PLAN.template.md
• assets/NEXT_SESSION_PLAN.template.md
• assets/IMPLEMENTATION_REPORT.template.md
• assets/project.template.json

Core Engine (advanced)

Direct engine script: skills/project-ops/scripts/project-ops.ps1

Actions:

• init
• list
• menu
• choose
• status
• continue
• update
• assign
• kickoff
• log
• archive
• suggest

Data Layout

.projects/
├── index.json
└── <project-id>/
    ├── project.json
    ├── MASTER_PLAN.md
    ├── NEXT_SESSION_PLAN.md
    ├── ARCHIVE.md (optional; created on archive)
    ├── logs/YYYY-MM-DD.md
    └── Sxx/IMPLEMENTATION_REPORT.md

Portable Packaging (.skill)

Package this skill for reuse on another OpenClaw host:

python "C:/Users/Administrator/AppData/Roaming/npm/node_modules/openclaw/skills/skill-creator/scripts/package_skill.py" "C:/Users/Administrator/.openclaw/workspace/skills/project-ops" "C:/Users/Administrator/.openclaw/workspace/dist"

Transfer dist/project-ops.skill to the target machine, then unpack into its workspace/skills/project-ops and restart gateway.

Standard Loop

1. /project
2. /project N
3. /project suggest
4. /project push <id>
5. Work execution + /project log <id> ...
6. /project archive <id> ... (when paused/completed)