agentleFS
Sign inStart free
sant1jmnzsant1jmnz/claude-code-kit/skills/weekly-update/SKILL.md

weekly-update

Skill2 starsChanged 3 days agoMITView on GitHub

Weekly check for updates of everything installed in the AI-agent setup (agent CLI, skills, plugins, MCP servers, RTK) and a short report. Read-only by default; applies nothing without the user's OK. Use when the scheduled weekly task fires, or when the user asks "check for updates", "is anything outdated", "update my skills/plugins".

8 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    After applying, tell the person to restart the agent.

    line 33
  2. Very specific:

    Never auto-update. Never run npx skills update with no names unless the user approved "everything".

    line 36
  3. Distinctive:

    Never use a model bigger than needed: this is a cheap, mostly mechanical job (see the cost rules in the global instructions).

    line 37
  4. Distinctive:

    Don't include secrets or personal data in the report.

    line 38
  5. Distinctive:

    If GitHub is unreachable (offline, rate-limited), say so and retry next week.

    line 39

“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “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

Weekly update check

Goal: keep the setup current without surprises. Third-party skills, plugins and MCP servers are code/instructions written by other people: updating them is a supply-chain decision, so this skill reports first and changes nothing until the user says yes.

Steps

  1. Run the checker (read-only, a few seconds, no model tokens spent reading files):

    python -I <this-skill-folder>/check-updates.py

    It reports: agent CLI version, skills whose source repo has newer commits, plugins whose marketplace has newer commits, version-pinned MCP packages with newer releases, and RTK. UPDATE = a newer version exists. UPDATE? = the repo has newer commits since you installed (they may not touch your skills). ok / skip need no action.

  2. Triage. For each UPDATE/UPDATE?, in one line: what, current → latest, and a risk note (pinned MCP major bump = review; skills from an unknown author = read the diff). Don't dump the raw output at the user.
  3. Write the report to tracking/updates/YYYY-MM-DD.md if the user has a tracking folder, else ./updates/YYYY-MM-DD.md. Keep it under 30 lines: summary first, then the list, then "what I'd update now / what I'd wait on".
  4. Ask once, with the recommendation first: "Update X, Y, Z now? (I'd hold W because ...)". Wait for an explicit yes.
  5. Apply only what was approved, then verify:
    • Skills: npx skills update <names> -g -y
    • Plugins (Claude Code): /plugin marketplace update <marketplace> then /plugin update <plugin>@<marketplace> (UI commands: give them to the person if you can't run them)
    • MCP pinned versions: bump the @x.y.z in the MCP's args (claude mcp remove <name> + claude mcp add ... with the new version), then claude mcp list
    • RTK: reinstall from its GitHub release; verify with rtk gain
    • Agent CLI: follow the agent's own update command After applying, tell the person to restart the agent.

Rules

  • Never auto-update. Never run npx skills update with no names unless the user approved "everything".
  • Never use a model bigger than needed: this is a cheap, mostly mechanical job (see the cost rules in the global instructions).
  • Don't include secrets or personal data in the report.
  • If GitHub is unreachable (offline, rate-limited), say so and retry next week; don't guess.
  • Apply the anti-slop rule to the report: concrete versions and dates, no filler.

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.

Check your setup, free