update-framework
Skill18 starsChanged 3 months agoMITView on GitHub
Pull the latest framework version from upstream, rebuild Docker if needed, rebuild the listener, and restart PM2. Use when the user says "update framework", "update", "pull updates", "upgrade", or asks if there's a new version.
- Reads credentials
- Installs packages
In this repository9 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- create-agentSkill
- doctorSkill
- logsSkill
- manage-known-usersSkill
- manage-routinesSkill
- manage-work-hoursSkill
- setupSkill
- setup-watcherSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
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
- could be automatic: a setting could enforce it
- Distinctive: line 10
Run from repo root.
- Distinctive: line 19
If no new commits, tell the user "Already up to date" and stop.
- Very specific: line 21
If commits exist, summarize them (subjects only, max 10 lines) and ask the user to confirm before proceeding.
- Distinctive: line 29
If output is non-empty, the user has local changes to framework code.
- Very specific: line 39
If --ff-only fails (their main has diverged from origin), surface it.
- Very specific: line 65
If the changes touched listener/package.json, npm install is required (not just ci).
≈ “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
Update Framework
User content (agents/, shared/, config/, .env) is never touched by git pull because it lives in directories the framework doesn't write to. This skill performs the predictable update sequence safely.
Run from repo root.
Step 1 — Show what's available
git fetch
git log HEAD..origin/main --onelineIf no new commits, tell the user "Already up to date" and stop.
If commits exist, summarize them (subjects only, max 10 lines) and ask the user to confirm before proceeding. Look for words like BREAKING: or MIGRATION: in commit messages — flag those prominently.
Step 2 — Verify clean working tree on framework dirs
git status --porcelain -- listener/ docker/ framework/ scripts/ templates/ .claude/skills/If output is non-empty, the user has local changes to framework code. Ask: do they want to (a) stash, (b) commit first, or (c) abort? Don't pull on top of dirty framework state — merge conflicts in framework code are confusing.
User content (agents/, shared/, config/, .env) appearing in git status is fine — the pull won't touch those.
Step 3 — Pull
git pull --ff-only origin mainIf --ff-only fails (their main has diverged from origin), surface it. Don't auto-merge; let the user decide.
Step 4 — Detect what needs rebuilding
# Compare what changed against the previous HEAD (saved by git pull's reflog)
git diff --name-only HEAD@{1} HEADDecide:
docker/orframework/skills/changed → rebuild imagelistener/changed → rebuild listener- Anything else → skip rebuilds
Step 5 — Rebuild image (if needed)
docker build -t ginnie-agent -f docker/Dockerfile .Step 6 — Rebuild listener (if needed)
cd listener && npm install && npm run build && cd ..If the changes touched listener/package.json, npm install is required (not just ci).
Step 7 — Restart PM2
pm2 restart ecosystem.config.cjs --update-env--update-env re-reads the launching shell's env (so a refreshed CLAUDE_CODE_OAUTH_TOKEN or TZ change takes effect).
Step 8 — Verify
Invoke the doctor skill. Report any new failures. If listener fails to start, surface the last 30 lines of pm2 logs ginnie-agents-listener --nostream.
Step 9 — CHANGELOG hint
Print the relevant CHANGELOG.md section for the new version range so the user knows what changed in their framework.
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.

