agentleFS
link7373link7373/agentic-sales-marketing/.claude/skills/health-check/SKILL.md

health-check

Skill1 starsChanged 3 months agoMITView on GitHub

Audit the team itself — stale knowledge files, dead scheduled tasks, unauthorized connectors, broken cross-references, expired approvals, pipeline rot. Read-only; produces a risk-ranked fix list. Run quarterly, after any long gap, or whenever something feels off.

21 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use testing.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • NVIDIASkill

    testing

    Run Python and Rust tests for physicsnemo-curator using uv + pytest with coverage reporting via pytest-cov, and cargo-nextest for Rust tests. Includes benchmark workflows with pytest-benchmark and criterion.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 16 edited versions we read keep these.

  • 1

    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:

    Rank and report: Every finding as: what's broken, what it costs if ignored, and the one-line fix (usually "run skill X").

    line 24
  2. Very specific:

    Read knowledge/README.md (the who-writes/who-reads map) and playbook.md §8 (conventions) — they're the specification this audits against.

    line 29

“Common” is measured: kept by at least 6 in 10 of the 16 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

/health-check — Is the Team Itself Healthy?

Inputs: the whole repository — knowledge/, .claude/agents/, .claude/skills/, standards/, templates/. Agents: Head of Growth leads; tools-operator verifies connections and schedules.

Every other skill works on the business. This one works on the team. Systems that record their own state drift quietly: a scheduled task stops firing, a connector's authorization lapses, a skill writes to a file nobody reads anymore. Nothing announces itself. This finds it.

This skill is read-only except for its own report and knowledge/pending.md. It proposes fixes; it doesn't apply them.

Steps

  1. Knowledge base state. For each top-level file in knowledge/: read its status header. Flag stub files that downstream skills require (e.g., an approved strategy referencing a stub channel plan), and active files untouched for over 60 days where the rhythm says they should be current (pipeline.md, content-calendar.md, results-history.md, budget.md).
  2. Loop integrity. Cross-check against the who-writes/who-reads map in knowledge/README.md: any file written by a skill but read by no one (a dead-end loop), and any file an agent claims to read that doesn't exist. Both are real defects — v1 shipped with social-proof.md written and never read.
  3. Reference check. Every Read first path in .claude/agents/*.md, every Inputs/Writeback path in .claude/skills/*/SKILL.md, every template path, and every skill named in CLAUDE.md's routing table — does the target exist (or is it a documented runtime-created path following the slug convention)? Report each break with its source file.
  4. Tool registry vs reality. For each Connected row: does the connector still respond (harmless read-only call via tools-operator)? For each needs auth row a skill depends on: is the business blocked on it? Verify the credential-location column is a location, never a value.
  5. Scheduled automation. Are the weekly /measure and monthly /review-monthly tasks registered and alive (mcp__scheduled-tasks__list_scheduled_tasks)? A rhythm nobody runs is a rhythm that doesn't exist. If they're missing, this is usually the highest-risk finding.
  6. Approvals hygiene. In knowledge/approvals-log.md: expired approvals still being relied on, standing approvals with no limit or expiry (an unbounded blank cheque is a gate failure), and any executed action in results-history.md with no traceable approval ref.
  7. Pipeline and pending rot. Entries in knowledge/pipeline.md with no dated next action or a next action past due; items in knowledge/pending.md parked over 14 days (the operator was asked and it fell through the floor).
  8. Rank and report. Every finding as: what's broken, what it costs if ignored, and the one-line fix (usually "run skill X"). Ranked by risk — things silently losing money or breaking compliance first, tidiness last. A clean run says so plainly in one line.

Before you start

  1. Required inputs: none. This skill runs against whatever exists, including a half-built team — that's precisely when it's useful.
  2. Read knowledge/README.md (the who-writes/who-reads map) and playbook.md §8 (conventions) — they're the specification this audits against.
  3. Gate check: none. Read-only.

If things go wrong

  • A tool can't be verified (unauthorized, offline) → report as unverified, not as broken; state which and why.
  • The repository has drifted from the conventions in a way that's deliberate → don't "fix" it silently; report it and let the operator confirm which is right.
  • Findings exceed what's fixable now → deliver the ranked list and recommend the top three; a 40-item list nobody acts on is worse than three that get done.

Outputs

A risk-ranked fix list with a one-line fix per finding, and a one-line overall verdict. Close with the report per templates/skill-report.md.

Writeback

The report to knowledge/results-history.md (dated, under "Health check"); unresolved operator-blocked items to knowledge/pending.md. No other writes — fixes are separate, deliberate runs of the skills this recommends.

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.