stack-updater
Skill1 starsChanged 33 days agoMITView on GitHub
Trigger: check updates, update my stack, update skills, update gentle-ai, outdated tools, actualizar mi stack. Report available updates for bootstrap tools and this repo's skills, apply them on approval.
In this repository13 other files it gives its agents.
- AGENTS.mdAGENTS.md
- capture-insightsSkill
- create-runbookSkill
- datadog-logs-analyzerSkill
- datadog-metrics-investigatorSkill
- datadog-report-publisherSkill
- datadog-service-healthSkill
- datadog-triageSkill
- k8s-error-analyzerSkill
- k8s-optimizer-costSkill
- meeting-notes-enhanceSkill
- runSkill
- second-opinionSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 0 are distinctive, and 7 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
7
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 18
Never run sudo or system-package commands without showing the command and getting approval.
- Very specific: line 19
Use only the updater matching the binary's source from assets/scan.sh.
- Very specific: line 20
Never invent a "latest" version.
- Very specific: line 21
Never touch project repositories.
- Very specific: line 23
Before installing anything, run ./setup.sh --list, ask the user which tools and families they want, then run ./setup.sh --only <ids> --yes.
- Very specific: line 41
Run bash assets/scan.sh (path relative to this skill), gentle-ai update when gentle-ai is present, and ./install-external.sh list from the …
- Very specific: line 46
Run gentle-ai doctor only if Gentle AI was upgraded.
≈ “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
Activation Contract
Load when the user asks whether anything installed by this repo can be updated: bootstrap tools (those listed by ./setup.sh --list tools), Gentle AI-managed tools, the skills in this repository, or external (third-party) skills installed from external/catalog.tsv. Also load when the user names a new tool or repo to track.
Hard Rules
- Audit is read-only. NEVER update before the user approves the exact list.
- Not-installed tools are NOT a problem: list them in ONE line ("Not installed: a, b") with no advice, no install offer, no follow-up question.
- Never run
sudoor system-package commands without showing the command and getting approval. - Use only the updater matching the binary's
sourcefromassets/scan.sh. Never mix managers. - Never invent a "latest" version; report
unknown. - Never touch project repositories. Write only to
~/.config/personal-os-skills/stack.yaml, and only when the user names a new item. - Stay terse: no installation suggestions, no unrelated tooling.
- Before installing anything, run
./setup.sh --list, ask the user which tools and families they want, then run./setup.sh --only <ids> --yes. Never runsetup.sh,install.shorbootstrap-harness.shwithout an explicit selection.
Decision Gates
| Situation | Action |
| Gentle AI, Engram, GGA | Use gentle-ai update for installed vs latest; apply with gentle-ai upgrade |
| Claude / Cursor / Codex | Source-specific command in references/update-commands.md |
| External skill installed | Status from ./install-external.sh list (run from the repo root); latest is unknown because the skills CLI has no check-only mode. Update only on approval: npx --yes skills@<SKILLS_CLI_VERSION> update -g <skill>, with the pin read from install-external.sh |
Repo behind > 0 | git pull --ff-only, then ./install.sh --no-prompt --family <families already installed> to re-render placeholder skills |
Repo dirty > 0 | Report it; do not pull until the user decides |
Skill missing for a detected harness | Report names in one line; install only via ./setup.sh --only <family> on approval |
Skill copy with placeholders | Re-run ./install.sh --no-prompt --family <its family> after a pull to refresh |
source = unknown | Report; do not guess a manager |
| Update fails | Stop that item, show output, continue the rest |
Execution Steps
- Run
bash assets/scan.sh(path relative to this skill),gentle-ai updatewhengentle-aiis present, and./install-external.sh listfrom the repo root. - Merge extra items from
~/.config/personal-os-skills/stack.yamlif it exists. - Build the report; consider cursor skills only if
~/.cursorexists. - Show it, then ask once: update all / pick / none. STOP and wait.
- Apply approved items one at a time; re-check each version.
- Run
gentle-ai doctoronly if Gentle AI was upgraded.
Output Contract
Table item | current | latest | status (up-to-date | update available | unknown), skills summary (N installed, M missing), one line for not-installed tools. After approval: item | command | result | new version, plus skipped and failed items.
References
references/update-commands.md— per-source commands.assets/scan.sh— read-only detector.assets/stack.example.yaml— manifest for extra items.
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.