week-report-system
Skill9 starsChanged 9 months agoMITView on GitHub
Material-based AI weekly report system. Use when the user mentions 周报, week report, weekly report, 本周总结, 工作总结, 写周报, 生成周报, 帮我写周报, 本周工作, 上周周报, 查看周报, 周报系统, 记录工作, 纳入周报素材, 周报素材, plan, experiment, 方案文档, 复盘文档, or asks to summarize recent repo activity into weekly report materials. This skill captures work materials, syncs them to Git, and generates weekly reports from structured materials.
In this repository15 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- alapiSkill
- android-app-publisherSkill
- awesome-docsSkill
- calib-pattern-generatorSkill
- deep-thinkSkill
- learning-tutorSkill
- proactive-agentSkill
- research-paper-deep-diveSkill
- send-emailSkill
- shell-shortcutsSkill
- technical-explainer-videoSkill
- tutor-generalSkill
- tutor-math-geometrySkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
18 rules. About 0 are common, 8 are distinctive, and 10 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
This author's own choices. Worth a look.
10
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 18
If the user explicitly says 纳入周报素材, 记入周报, 作为周报素材, or equivalent, you must execute material capture.
- Very specific: line 19
When generating a weekly report, you must check {year}/week{WW}/materials/ first.
- Very specific: line 20
If any readable materials/*.jsonl files exist, the main report analysis must be based on them.
- Distinctive: line 23
If Git sync fails after a capture attempt, you must preserve local status evidence.
- Distinctive: line 24
If one session crosses midnight, captured materials must be split by calendar date in file naming.
- Distinctive: line 31
Do not claim guaranteed logging for every conversation
- Very specific: line 32
When the user cares about completeness, explicitly ingest materials via scripts/material_ingestor.py
- Distinctive: line 36
Always run this check first:
- Distinctive: line 44
If all three are set, continue with the request
- Very specific: line 45
If any is missing, read references/setup_guide.md and guide the user through setup
- Very specific: line 64
Use scripts/material_ingestor.py or scripts/conversation_logger.py to persist the event
- Very specific: line 65
Treat materials/*.jsonl as the source of truth
- Very specific: line 66
Write *.txt digest only as a compatibility/debug view
- Distinctive: line 67
Update local sync status files for observability
- Very specific: line 72
Read references/report_generation.md for the full process.
- Very specific: line 76
Check materials/*.jsonl under {year}/week{WW}/ first
- Distinctive: line 77
If any structured materials exist, base the report on them
- Distinctive: line 103
Prefer explicit material ingestion for important plans, experiments, or repo analysis
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Week Report System
Material-first weekly report workflow for work conversations, documents, and recent repository activity.
What This Skill Does
- Capture work materials into a Git-backed weekly knowledge base
- Support three material types:
conversation,document,repo_activity - Generate weekly reports from structured materials, with legacy conversation logs as fallback
Hard Rules
- If the user explicitly says
纳入周报素材,记入周报,作为周报素材, or equivalent, you must execute material capture. - When generating a weekly report, you must check
{year}/week{WW}/materials/first. - If any readable
materials/*.jsonlfiles exist, the main report analysis must be based on them. - You must not generate a report only from
*.txtwhen structured materials are present. *.txtdigests are a compatibility/debug artifact, not the source of truth.- If Git sync fails after a capture attempt, you must preserve local status evidence.
- If one session crosses midnight, captured materials must be split by calendar date in file naming.
Important Boundary
This skill can do best-effort automatic capture only when the host agent actually invokes it.
- A skill description is not a system-level post-turn hook
- Do not claim guaranteed logging for every conversation
- When the user cares about completeness, explicitly ingest materials via
scripts/material_ingestor.py
Step 0: First Time — Check Environment
Always run this check first:
echo "USERNAME: ${WEEK_REPORT_GIT_USERNAME:-MISSING}"
echo "TOKEN: $([ -n \"$WEEK_REPORT_GIT_PERSONAL_TOKEN\" ] && echo 'SET' || echo 'MISSING')"
echo "REPO: ${WEEK_REPORT_GIT_REPO:-MISSING}"- If all three are set, continue with the request
- If any is missing, read
references/setup_guide.mdand guide the user through setup
Determine User Intent
| User intent | Action |
| "写周报" / "生成周报" / "本周总结" / "上周周报" | Generate report → references/report_generation.md |
| "把这个纳入周报素材" / "记录这次讨论" | Must capture as material → references/material_ingestion.md |
| "总结这个 plan/experiment 文档并纳入周报" | Ingest as document material |
| "看某个 repo 最近几天提交并纳入周报" | Ingest as repo_activity material |
| "周报系统怎么用" / "怎么设置" / "skill介绍" | Show guide → references/user_guide.md |
| Any other work-related message | Answer normally, then best-effort capture a conversation material |
Material Capture Workflow
Full instructions: references/material_ingestion.md
Quick summary:
- Build a structured material event with
source_type,project,title,summary,evidence,tags,outcome,next_actions - Use
scripts/material_ingestor.pyorscripts/conversation_logger.pyto persist the event - Treat
materials/*.jsonlas the source of truth - Write
*.txtdigest only as a compatibility/debug view - Update local sync status files for observability
- Fail silently for the user, but leave local status/queue evidence for debugging
Weekly Report Generation
Read references/report_generation.md for the full process.
Quick summary:
- Pull latest data from Git
- Check
materials/*.jsonlunder{year}/week{WW}/first - If any structured materials exist, base the report on them
- Fall back to legacy
.txtlogs only if structured materials are missing or unreadable - Group by project / workstream, extract evidence and outcomes, then render the report
- Optionally save the report to
{year}/week{WW}/report-{YYYYMMDD}-{HHmmss}.md
Reference Files
| File | When to read |
references/setup_guide.md | Missing env vars |
references/material_ingestion.md | Capturing conversation, document, or repo materials |
references/conversation_tracking.md | Best-effort conversation capture and local observability |
references/report_generation.md | Generating reports |
references/user_guide.md | User asks how to use the system |
Scripts
| Script | Purpose |
scripts/git_operations.py | Clone / pull / push the report repo |
scripts/conversation_logger.py | Capture conversation materials and maintain digest files |
scripts/material_ingestor.py | Explicitly ingest conversation, document, or repo_activity materials |
Important Notes
- Best-effort auto capture is useful, but not a guarantee of complete logging
- Prefer explicit material ingestion for important plans, experiments, or repo analysis
- Skip recording only for high-confidence sensitive content such as passwords, private keys, PATs, or explicit "不要记录"
- Recording failures must not interrupt the user response, but they should leave status clues locally
- Structured material events live in
materials/*.jsonland are the primary source for report generation - One session can append to one digest file per calendar date for compatibility/debugging
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

