Project Milestone Tracker 项目管理里程碑跟踪器
SkillAI & modelsLets your agent plan and track project milestones and generate status reports.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the Project Milestone Tracker 项目管理里程碑跟踪器 skill
About this capability
项目管理里程碑跟踪器(Project Milestone Tracker)——当用户要求代理在代码库/项目会话中规划、跟踪里程碑、记录承诺或事件、生成进度报告时使用。核心:三层里程碑规则(文件依据/客户提供/待确认)+ 原话+时间戳事件记录。Project milestone tracker: plan and track project milestones, record events with original text + timestamp, generate status reports, using a three-tier milestone rule (contract-so
What this skill tells your AI
The instructions your AI receives, as published by davepoon/buildwithclaude in plugins/all-skills/skills/project-milestone-tracker/SKILL.md and read by ahel’s review.
A self-contained local tracker for planning and tracking project work in a repo session — no external service, no cloud, pure local files (Python stdlib only).
When to Use
Use this skill when the user asks the agent to:
- Plan milestones for a project / break work into tracked milestones
- Track milestone progress and mark items done
- Record a promise, decision, or event with original text + timestamp (traceable claims)
- Generate a project status report (markdown)
It is designed for repo-session work: the agent runs local commands and keeps state in a local JSON file.
Three-Tier Milestone Rule (三层里程碑规则)
Never invent milestone dates. Every milestone must carry a source tier:
| Tier | source | Meaning | Behavior |
|---|---|---|---|
| 1 | file | Extracted from a contract/document | Authoritative; note the source document |
| 2 | client | Stated verbally by stakeholder | Recorded as stated; suggest written confirmation |
| 3 | pending | No milestone info yet | Flag "ask the client" — never self-plan |
How to Use
All commands run via the bundled script (stdlib only, no install):
python3 scripts/pm_track.py --file .pm.json init "东区改造项目"
python3 scripts/pm_track.py --file .pm.json add-milestone "3月15日竣工验收" --source file --due 2026-03-15 --note "依据合同第四章"
python3 scripts/pm_track.py --file .pm.json add-milestone "甲方口头承诺月底回款" --source client --due 2026-03-31
python3 scripts/pm_track.py --file .pm.json add-milestone "材料进场节点待确认" --source pending
python3 scripts/pm_track.py --file .pm.json add-event "甲方说马上回款" --who 王总 --when "2026-03-10 14:30"
python3 scripts/pm_track.py --file .pm.json status
python3 scripts/pm_track.py --file .pm.json report # markdown report
python3 scripts/pm_track.py --file .pm.json done 1
Commands
| Command | What it does |
|---|---|
init <name> | Initialize a project tracker file |
add-milestone <title> --source file|client|pending [--due] [--note] | Add a milestone with a source tier |
add-event <text> [--who] [--when] | Record an event with original text + timestamp |
done <id> | Mark a milestone done |
status | Show all milestones + events + pending warnings |
report | Print a markdown progress report |
Example
$ python3 scripts/pm_track.py --file .pm.json init "门店装修"
$ python3 scripts/pm_track.py --file .pm.json add-milestone "6月10日完工验收" --source file --due 2026-06-10
$ python3 scripts/pm_track.py --file .pm.json add-milestone "水电改造完成" --source client
$ python3 scripts/pm_track.py --file .pm.json add-event "甲方电话说马上回款" --who 李总
$ python3 scripts/pm_track.py --file .pm.json status
📋 Project: 门店装修
=== 里程碑 (2 open) ===
⬜ #1 [file ] 6月10日完工验收 (due 2026-06-10)
⬜ #2 [client] 水电改造完成
=== 事件记录 (1) ===
• [2026-08-25 15:20] (李总) 甲方电话说马上回款
Data & Portability
- State is stored in a local JSON file (default
.project-tracker.json, or pass--file). - Pure Python 3.8+ stdlib (
argparse,json,datetime) — no pip installs, no network. - Works anywhere the agent has a shell. The JSON file can be committed to the repo or git-ignored.
Templates
A milestone planning checklist and a status report template live in templates/ — copy and adapt.
Pitfalls
- Never invent due dates for
pendingmilestones — the whole point of the three-tier rule is that the agent asks the client instead of guessing. - Store original text, not paraphrases — events keep
textverbatim pluswhen/whoso claims are traceable. reportprints markdown to stdout; pipe to a file (> report.md) to save.
Signals
- GitHub stars
- 3k
- Forks
- 500
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
project-milestone-tracker- Source
- github.com/davepoon/buildwithclaude