Use this skill when the user explicitly asks to set up, resume, update, or hand off persistent project context for AI-assisted work. Triggers (all require an explicit user request — do NOT activate just because a brain/ folder exists): (1) The user wants to set up project-level memory ("set up project brain", "scaffold project context", "init project brain", "建项目脑"). (2) The user is in a directory containing brain/ AND explicitly asks to resume / continue / load the project / check status / "what's the state of this project". (3) The user signals window switch / context compaction ("switch windows", "context's getting full", "I'll head out", "压缩了") — write a HANDOFF before they leave. (4) The user says "update the project brain" / "let's record this" / "更新项目脑" — propose a list with reasons; user approves per item. Do NOT auto-activate during casual conversation that happens to occur inside a brain/-bearing directory. The methodology's "activation boundary" requires an explicit user request. This skill provides a folder structure (brain/), copy-able templates for multiple AI tools, and a four-workflow protocol (startup / kick-off / update / handoff). Methodology details in METHODOLOGY.md alongside this file.
Resources
9Install
npx skillscat add ethan-ys/project-brain Install via the SkillsCat registry.
project-brain
A folder structure + collaboration protocol for persistent project context across AI sessions. Use it to scaffold a brain/ folder in the user's project, resume work from an existing one, or update entries when work happens.
When to invoke this skill
| User signal | Workflow |
|---|---|
| "set up project brain" / "scaffold project" / "建项目脑" / "init project context" | Kick-off — see §1 |
User is in a directory containing brain/ AND explicitly says "load" / "resume" / "what's the status" / "继续这个项目" |
Startup — see §2 |
| User says "switch windows" / "context's getting full" / "I'll head out" / "压缩了" | Handoff — see §3 |
| User says "update the project brain" / "let's record this" / "更新项目脑" | Update — see §4 |
Activation boundary (important): Do not activate just because a brain/ folder exists in the current directory. Casual conversation that happens to occur inside a project does not require this skill — only explicit user requests do. This is intentional: the methodology refuses to auto-monitor or auto-trigger; user judgment decides when to engage.
Core principles (always honored)
- Don't auto-modify any file in
brain/. Always propose; user approves. - Judgment Division: the user decides "should we record now"; the AI decides "what specifically to record"; the user reviews. Don't push specialized judgments back to the user ("which files do you want me to update?").
- DECISIONS gentle inquiry: when something feels decided, ask "does this count as decided?" — don't assert "we decided X."
- Multi-workstream: if
brain/has multipleSTATUS_<workstream>.mdfiles, don't guess this window's workstream — ask the user.
§1 New project kick-off
When the user wants to scaffold:
- Confirm applicability: multi-module / long-lived / cross-session work? If a one-off script — say "this might be over-engineering for this project" and let the user override.
- Confirm single or multi-workstream: most projects are single-workstream (default). If the project has parallel independent workstreams (dev + ops + outreach kind of project), ask the user to list workstream names.
- Run scaffold:
bash <skill-path>/scripts/scaffold.sh <user-project-root>(defaults to all four AI adapters: Claude, Cursor, Copilot, AGENTS.md). - Walk the user through
brain/PROJECT.mdon day one — fill the one-line definition + "what we explicitly DON'T do." Don't let this drift into the future. - Scan placeholders:
grep -rn "⚠️ TODO ⚠️" <user-project>/brain/— go through with the user. Empty fields should be intentional. - First DECISIONS entry: append "establishing project-brain" with rejected alternatives, so this file is used from day one.
§2 Startup (entering a project that already has brain/)
Single-workstream (only STATUS.md):
- Read
brain/MAP.md - Read
brain/STATUS.md - If
brain/HANDOFF.mdexists, read it - Brief report: project recognized + current progress + last blocker
Multi-workstream (multiple STATUS_<workstream>.md):
- Read shared files:
brain/MAP.md+brain/PROJECT.md - Don't guess — ask: "I see this is a multi-workstream project with [list workstreams]. Which one does this window work on?"
- After the user answers, read corresponding
brain/STATUS_<workstream>.md+brain/HANDOFF_<workstream>.md - Brief report
Universal discipline: if cwd switches to another project mid-session, re-read the new project's files; if the user switches workstreams in the same window, re-read the new workstream's files. Don't carry over memory.
§3 Handoff (user signaling window switch)
When the user says they're switching windows / context is getting full / heading out:
- Archive the existing HANDOFF first (if one exists):
(For multi-workstream:git mv brain/HANDOFF.md brain/handoffs/$(date +%Y-%m-%d-%H%M).mdbrain/HANDOFF_<current>.md→brain/handoffs/<current>/....) - Write a fresh
brain/HANDOFF.mdcapturing:- Why the window-switch (context limit / break / refocus)
- Current state at this exact moment
- Where the next session should pick up
- Things still in head not yet written down — the unique value of HANDOFF (hunches, half-tried approaches, weird debugging observations)
- Keep it short. If STATUS was just overwritten, HANDOFF can be near-empty.
§4 Update workflow
When the user says "update the project brain":
- Look back at what happened in this session
- Propose a list with reasons — for each candidate file, say what to write and why:
- STATUS overwrite: now at [X], next [Y]. Reason: you said "that's it" - DECISIONS: about to append? Reason: that section felt like it landed — does this count as decided? - MAP §5 register: new file at brain/topics/systems/[X].md. Reason: doc we just wrote - HANDOFF: not needed — you didn't say switch windows - Wait for user's approval per item — they may say "OK, all of them" / "skip DECISIONS, that's still discussion" / "today, just STATUS"
- Write only the approved updates
Never reverse this: don't ask the user "which files do you want to update?" That hands the wrong layer of judgment to the wrong person.
Reference
- Full methodology (the why, all 14 traps, evolution story):
METHODOLOGY.md(alongside this file) - Templates:
templates/(whatscaffold.shcopies) - Public repo: https://github.com/Ethan-YS/project-brain