agentleFS
aka-kikaaka-kika/akakika-skills/skills/project-intelligence/project-catch-up/SKILL.md

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.

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
  1. Very specific:

    Read PROJECT_GUIDE.md if present.

    line 22
  2. Distinctive:

    Write the report from the template.

    line 25
  3. Distinctive:

    Prefer official sources: release notes, migration guides, docs, GitHub releases.

    line 37
  4. Distinctive:

    Run targeted searches (not one vague query).

    line 53
  5. Distinctive:

    Use the WebSearch tool (or equivalent).

    line 70
  6. Distinctive:

    Use the answer to set the research window and tone ("you left off mid-auth refactor").

    line 79
  7. Very specific:

    If Pieces fails, use git log -1 --format=%ci when the repo is git—mention the date in the report.

    line 81
  8. Distinctive:

    Do not list 20 tools—curate to what matches this repo

    line 166
  9. Distinctive:

    If nothing important changed, say so clearly: *You're not missing much—safe to continue.*

    line 167
  10. Distinctive:

    Do not recommend upgrading everything "just because"

    line 172
  11. Distinctive:

    Do not invent version numbers—read the repo or say unknown

    line 173
  12. Distinctive:

    Do not skip web research

    line 174
  13. Distinctive:

    Do not ask whether to save the file

    line 175
  14. Very specific:

    Do not replace PROJECT_GUIDE.md or PROJECT_VERDICT.md

    line 176

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Project Catch-Up

Where this fits

Part of the same workflow as:

  1. explain-new-project → PROJECT_GUIDE.md
  2. project-worth-my-time → PROJECT_VERDICT.md (optional)
  3. 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

  1. Confirm project root.
  2. Read PROJECT_GUIDE.md if present; else README + lockfile/manifest to list the stack (frameworks, major packages, hosting, AI tools).
  3. Pieces MCP (try once) — when did the user last touch this project? (see below). Skip judgment if unavailable.
  4. Web research (required) — search for news since a sensible window (see Research window).
  5. Write the report from the template; cite sources with links where possible.
  6. Always save PROJECT_CATCH_UP.md at the project root (overwrite if present).
  7. In chat: show Before you code + Headline in full; one line points to the file for the full list.

Research window

SituationSearch focus
User says how long they were awayChanges since that period
Pieces shows last activity dateRoughly from that date to today
UnknownLast 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 repo

Then 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:

  1. get_user_persona() — optional context for how they learn.
  2. 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:

LabelMeaning
Must knowBreaking change, security fix, or will waste your time if ignored
Worth a lookNew capability that could replace work you planned
Nice to haveOptional polish, new skill, minor version bump
Ignore for nowIrrelevant 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.md or PROJECT_VERDICT.md

Multi-root workspaces

One PROJECT_CATCH_UP.md per named project root.


Examples: examples.md

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.