跳至正文
OpenAgentSkill
浏览已开放 · 目录记录日期 2026-11-21 · 自动更新和账户服务暂时关闭。
搜索与研究

worklog-analysis

Analyze past work logged by the worklog-logging skill. Generates daily standups, weekly summaries, monthly reviews, and performance review material. TRIGGER THIS SKILL when the user says "standup", "what did I do", "weekly summary", "monthly review", "performance review", "resume update", "what have I been working on", or asks to analyze, summarize, or review past work over any timeframe. This skill reads worklog files — it does NOT log new entries. For logging, use worklog-logging.

thumperL · 已有静态检查记录

安装与使用

运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。

npx skills add thumperL/claude-worktrace --skill worklog-analysis

技能说明


name: worklog-analysis description: > Analyze past work logged by the worklog-logging skill. Generates daily standups, weekly summaries, monthly reviews, and performance review material. TRIGGER THIS SKILL when the user says "standup", "what did I do", "weekly summary", "monthly review", "performance review", "resume update", "what have I been working on", or asks to analyze, summarize, or review past work over any timeframe. This skill reads worklog files — it does NOT log new entries. For logging, use worklog-logging.

Worklog Analysis

Read past worklog entries and synthesize them for standups, reviews, and career documentation.

Where worklogs live

~/Documents/AI/worklog/
├── 2026-03-08-macbook-pro.md
├── 2026-03-07-mac-mini.md
└── ...

Naming: YYYY-MM-DD-{hostname}.md

Fallback: ~/.claude/worklog/

Use the bundled script for all queries, or read files directly.

Analysis commands

Daily standup

Trigger: "standup", "what did I do yesterday", "daily update"

python "${CLAUDE_PLUGIN_ROOT}/scripts/analyze_worklog.py" --standup

Output format:

🧍 Standup — March 8, 2026

**Yesterday:**
- [Items grouped by project, with session IDs for context]

**Today (carry-over):**
- [Open items from yesterday]

**Blockers:** None

When presenting a standup, keep it tight — this is what the user will paste into Slack or read in a meeting. No fluff.

Weekly summary

Trigger: "weekly summary", "what did I do this week"

python "${CLAUDE_PLUGIN_ROOT}/scripts/analyze_worklog.py" --weekly

Output format:

📊 Week of March 2–8, 2026

**Projects:** [list]
**Sessions:** [count across machines]

**Key accomplishments:**
[Grouped by project, most significant first]

**Technologies:** [aggregated]
**Decisions:** [notable ones]
**Artifacts:** [PRs, deployments, docs]

Monthly review

Trigger: "monthly review", "performance review", "what did I do last month"

python "${CLAUDE_PLUGIN_ROOT}/scripts/analyze_worklog.py" --monthly

Output format:

📈 Monthly Review — February 2026

**Overview:** [1-2 sentences — scope, impact, themes]

**By project:**
- [Project]: [Summary, outcomes, metrics]

**Skills & technologies:** [aggregated — for resume]

**Key achievements:** [Top 3-5 by impact]

**Patterns:** [Time allocation, recurring themes, growth areas]

**Resume-ready bullets:**
- [Impact-oriented achievement with quantification]
- [Another, framed for performance reviews]

The resume-ready bullets translate raw work into impact language. Quantify where possible: "Reduced API latency by 40% by implementing Redis caching layer" not "Improved performance".

Custom queries

Trigger: "what did I do on [date]", "show me [timeframe]"

python "${CLAUDE_PLUGIN_ROOT}/scripts/analyze_worklog.py" --query "2026-03-05"
python "${CLAUDE_PLUGIN_ROOT}/scripts/analyze_worklog.py" --query "this-week"
python "${CLAUDE_PLUGIN_ROOT}/scripts/analyze_worklog.py" --query "2026-03"  # whole month

Session tracing

Trigger: "what happened in session sess-f3a1", "trace that session"

When the user asks about a specific session, grep the worklog files for that session ID and reconstruct the full timeline of what happened in that session across all its log entries.

python "${CLAUDE_PLUGIN_ROOT}/scripts/analyze_worklog.py" --session "sess-f3a1"

Session grouping

Multi-segment sessions (same sess-XXXX across PreCompact + SessionEnd entries) are automatically grouped into single entries in standup, weekly, and monthly views. Summary bullets are deduplicated within grouped entries.

Individual segments are still visible via --session sess-XXXX trace, which shows the full timeline of all entries for that session.

Analysis tips

When generating reviews (weekly, monthly):

  • Read ALL relevant files before synthesizing — don't summarize from a subset
  • Cross-reference across machines to build the complete picture
  • Deduplicate items that appear in multiple entries (same work logged at different points)
  • Highlight decisions and their rationale — these are gold for performance reviews
  • Frame achievements by impact, not effort ("Shipped X that enabled Y" not "Spent 3 days on X")