project-catch-up
Skill11 starsChanged 3 months agoMITView on GitHub
Use when returning to a repo after a break, asking what changed, what's new for this stack, or invoking project-catch-up after explain-new-project. Scans the stack for important updates, deprecations, new helper tools, agent skills, and MCP servers online—plain-language brief saved as PROJECT_CATCH_UP.md.
In this repository39 other files it gives its agents.
- agent-starter-packSkill
- agent-state-machineSkill
- local-ai-ollamaSkill
- apple-hig-command-paletteSkill
- apple-hig-empty-statesSkill
- apple-hig-feedback-statusSkill
- apple-hig-inspectorsSkill
- apple-hig-macos-window-layoutSkill
- apple-hig-search-filteringSkill
- apple-hig-settingsSkill
- apple-hig-sf-symbolsSkill
- apple-hig-sidebarsSkill
- apple-hig-toolbarsSkill
- apple-hig-typographySkill
- blog-post-publishingSkill
- social-post-kitSkill
- color-principlesSkill
- kika-design-systemSkill
- emergency-switch-appSkill
- product-hunt-polish-reviewSkill
- docs-updateSkill
- project-folder-cleanupSkill
- between-runs-auditSkill
- explain-new-projectSkill
- new-project-gateSkill
- old-project-auditSkill
- project-workflowSkill
- project-worth-my-timeSkill
- butterfly-effectSkill
- spreadSkill
- tilSkill
- apple-app-intentsSkill
- macos-app-distribution-dmgSkill
- macos-clipboard-pasteboardSkill
- macos-global-shortcutsSkill
- macos-launch-at-loginSkill
- macos-menubar-swiftuiSkill
- macos-notificationsSkill
- macos-permissions-privacySkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
14 rules. About 0 are common, 11 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈11
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 22
Read PROJECT_GUIDE.md if present.
- Distinctive: line 25
Write the report from the template.
- Distinctive: line 37
Prefer official sources: release notes, migration guides, docs, GitHub releases.
- Distinctive: line 53
Run targeted searches (not one vague query).
- Distinctive: line 70
Use the WebSearch tool (or equivalent).
- Distinctive: line 79
Use the answer to set the research window and tone ("you left off mid-auth refactor").
- Very specific: line 81
If Pieces fails, use git log -1 --format=%ci when the repo is git—mention the date in the report.
- Distinctive: line 166
Do not list 20 tools—curate to what matches this repo
- Distinctive: line 167
If nothing important changed, say so clearly: *You're not missing much—safe to continue.*
- Distinctive: line 172
Do not recommend upgrading everything "just because"
- Distinctive: line 173
Do not invent version numbers—read the repo or say unknown
- Distinctive: line 174
Do not skip web research
- Distinctive: line 175
Do not ask whether to save the file
- Very specific: line 176
Do not replace PROJECT_GUIDE.md or PROJECT_VERDICT.md
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
Project Catch-Up
Where this fits
Part of the same workflow as:
- explain-new-project →
PROJECT_GUIDE.md - project-worth-my-time →
PROJECT_VERDICT.md(optional) - This skill →
PROJECT_CATCH_UP.md— run when you come back to a project so you do not rebuild what already exists or miss breaking updates
Same audience as explain-new-project: self-taught, plain language, Learn: for jargon, language matches the user's chat.
When invoked
- Confirm project root.
- Read
PROJECT_GUIDE.mdif present; else README + lockfile/manifest to list the stack (frameworks, major packages, hosting, AI tools). - Pieces MCP (try once) — when did the user last touch this project? (see below). Skip judgment if unavailable.
- Web research (required) — search for news since a sensible window (see Research window).
- Write the report from the template; cite sources with links where possible.
- Always save
PROJECT_CATCH_UP.mdat the project root (overwrite if present). - In chat: show Before you code + Headline in full; one line points to the file for the full list.
Research window
| Situation | Search focus |
| User says how long they were away | Changes since that period |
| Pieces shows last activity date | Roughly from that date to today |
| Unknown | Last 6 months for major stack items; 12 months for agent skills / MCP ecosystem |
Prefer official sources: release notes, migration guides, docs, GitHub releases. Avoid rumor blogs.
Stack inventory (from the repo)
Extract a short list before searching—example fields:
- Runtime: Node 22, Python 3.12, Swift 6
- Framework: Next.js 15, Vite, Expo, SwiftUI
- Data/auth: Supabase, Prisma, Firebase
- Deploy: Vercel, Cloudflare
- AI/agent: coding agent in use, specific MCP servers in the agent's config (e.g.
.cursor/,.codex/,.mcp.json) or docs
Also note pinned versions in lockfiles when visible (helps spot "you're on old X").
Web research checklist
Run targeted searches (not one vague query). For each major stack item, check:
- [ ] Latest stable release vs project's version — breaking changes?
- [ ] Deprecated APIs or "migrate before date" warnings
- [ ] Official replacement or recommended upgrade path
- [ ] New first-party features that replace custom code in this repoThen search the workflow layer:
- [ ] Agent skills: new or popular skills for [framework] / [task]
- [ ] MCP servers that match this project (official + well-maintained community)
- [ ] CLI or SaaS tools that fit the same job (only if clearly better than reinventing)Use the WebSearch tool (or equivalent). Run multiple searches; merge duplicates.
Pieces MCP (optional but valuable)
Via user-pieces:
get_user_persona()— optional context for how they learn.ask_pieces_ltm— When did I last work on [project name or folder path]? What was I doing?- Topics: project name, repo path, last session
- Use the answer to set the research window and tone ("you left off mid-auth refactor").
If Pieces fails, use git log -1 --format=%ci when the repo is git—mention the date in the report.
Severity labels
Tag each finding so the user can skim:
| Label | Meaning |
| Must know | Breaking change, security fix, or will waste your time if ignored |
| Worth a look | New capability that could replace work you planned |
| Nice to have | Optional polish, new skill, minor version bump |
| Ignore for now | Irrelevant to this repo or too bleeding-edge |
Report template
# [Project name] — catch-up before you code
> **Before you code** — read the headline and action list, then stop if you're ready.
> **The rest** — updates, tools, skills, and links.
---
## Before you code
**Headline:** [One sentence: safe to continue / pause for upgrade / big opportunity you almost missed]
**Do these first (if any):**
1. [Ordered, concrete—e.g. run npm update, read migration link, install MCP X]
2. ...
**You can ignore for now:** [Short list or "nothing critical"]
---
## The rest (optional)
### Your stack snapshot
| Piece | In this project | Note |
|-------|-----------------|------|
| ... | version or "unknown" | ... |
### Must know
| What | Why it matters to you | Source |
|------|----------------------|--------|
| ... | plain language | link |
### Worth a look — updates & features
[Framework/library releases, new APIs, hosting changes]
### Worth a look — tools & integrations
[CLIs, services, templates—not full app replacements unless relevant]
### Worth a look — agent skills & MCP
| Name | What it helps with | How to try |
|------|-------------------|------------|
| ... | ... | install path or search term |
**Learn:** [only if "MCP" or "skill" needs explaining]
### Nice to have
[Bullet list]
### Ignore for now
[Bullet list]
### When you last worked on this
[From Pieces or git date + one line of context]
### Sources checked
- [dated list of main URLs searched]Tone
- Goal: save time, not FOMO
- Do not list 20 tools—curate to what matches this repo
- If nothing important changed, say so clearly: You're not missing much—safe to continue.
- Distinguish upgrade the project vs use a new external tool vs add an agent skill
What not to do
- Do not recommend upgrading everything "just because"
- Do not invent version numbers—read the repo or say unknown
- Do not skip web research
- Do not ask whether to save the file
- Do not replace
PROJECT_GUIDE.mdorPROJECT_VERDICT.md
Multi-root workspaces
One PROJECT_CATCH_UP.md per named project root.
Examples: examples.md
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.

