agentleFS
jackchukajackchuka/skills/claude-skill-prereq-audit/SKILL.md

claude-skill-prereq-audit

Skill15 starsChanged 9 days agoMITView on GitHub

Scan skills for prerequisite tools, MCP servers, and auth requirements, then check if everything is installed and authenticated. Offers to fix issues. Use when setting up a new machine, after installing skills, or to verify your environment. Triggers: "check prerequisites", "skill prereqs", "are my tools installed", "verify skill dependencies", "/claude-skill-prereq-audit".

  • Installs packages

The file

Show its rules and text

We split this skill into the rules it gives an agent

17 rules. About 0 are common, 7 are distinctive, and 10 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 10

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    After installing new skills

    line 27
  2. Distinctive:

    If user specified skill names, filter the list to matching directory names

    line 44
  3. Very specific:

    Read each SKILL.md file

    line 46
  4. Very specific:

    Run checks for all discovered dependencies.

    line 78
  5. Very specific:

    Run command -v <tool> — if exit code 0, installed.

    line 82
  6. Very specific:

    Use ToolSearch with query +<server> to check if any mcp__<server>__* tools are available in the session

    line 87
  7. Very specific:

    If tools found → installed = "yes".

    line 88
  8. Distinctive:

    If no tools found → installed = "no", auth = "—"

    line 89
  9. Very specific:

    Run gh extension list once (cache the result)

    line 93
  10. Very specific:

    Check if each extension name appears as a suffix in any line of the output (e.g., the line contains /<extension-name> or ends with the …

    line 94
  11. Distinctive:

    If no issues found, print "All prerequisites satisfied." and stop.

    line 119
  12. Distinctive:

    If issues exist, ask: "Want me to fix any of these?

    line 123
  13. Distinctive:

    After each fix attempt, re-run the check for that item and show the updated row.

    line 134
  14. Distinctive:

    When all selected fixes are attempted, re-print the full table with updated results.

    line 136
  15. Very specific:

    If ~/.claude/skills/ doesn't exist, report "No skills directory found at ~/.claude/skills/" and check only .claude/skills/

    line 140
  16. Distinctive:

    If a tool's auth check hangs (>10s), kill it and mark as "TIMEOUT"

    line 141
  17. Very specific:

    If brew is not installed, suggest manual install instructions instead

    line 142

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Skill Prerequisite Check

Scan installed skills, discover their tool/MCP/auth dependencies via heuristic parsing, check if everything is installed and authenticated, and offer to fix issues interactively.

When to Use

  • Setting up a new machine
  • After installing new skills
  • Verifying environment is ready for all skills
  • Debugging "tool not found" errors during skill execution

Arguments

Parse from the user's invocation:

  • Skill names: Any tokens after the command → filter to those skills only
  • Default: scan all skills in ~/.claude/skills/ and .claude/skills/

Workflow

Phase 1: Discover Skills

  1. Glob for ~/.claude/skills/*/SKILL.md and .claude/skills/*/SKILL.md
  2. Follow symlinks
  3. If user specified skill names, filter the list to matching directory names
  4. Exclude this skill (skill-prereq-check) from the scan list — its own SKILL.md contains tool references as parsing examples, not actual dependencies
  5. Read each SKILL.md file

Phase 2: Parse Prerequisites

For each SKILL.md, extract dependencies using these heuristics:

CLI tools:

  • Extract command names (first token of each line) from fenced code blocks marked as bash, shell, or sh
  • Also scan ## Prerequisites sections and prose for backtick-quoted tool names followed by "CLI", "installed", or "required"
  • Ignore shell builtins and coreutils: echo, cat, ls, mkdir, cd, grep, sed, awk, rm, cp, mv, date, wc, head, tail, sort, uniq, export, set, test, true, false, if, then, else, fi, for, do, done, while, case, esac, command, which, local, return, read, shift, printf, declare
  • Ignore git subcommands: the tool is git, not git commit
  • Deduplicate per skill

MCP servers:

  • Scan for mcp__<server>__ patterns — extract <server> as the MCP server name
  • Also scan for <Name> MCP server or <Name> MCP tools in prose — extract <Name> lowercased
  • Deduplicate per skill

gh extensions:

  • Scan for gh <word> where <word> is NOT a built-in gh command: pr, issue, release, api, auth, repo, project, run, extension, gist, ssh-key, gpg-key, secret, variable, codespace, label, ruleset, cache, attestation, status, browse, search, workflow, config, alias
  • Record as gh extension dependency

Auth hints:

  • Scan for auth status, auth login, auth test, auth check, authenticated within 200 characters of a tool name
  • Mark that tool as needing an auth check

Phase 3: Check Dependencies

Run checks for all discovered dependencies. Use parallel Bash calls where possible. Deduplicate check commands across skills — if multiple skills depend on the same tool, run the check once and reuse the result.

CLI tools:

  1. Run command -v <tool> — if exit code 0, installed; otherwise missing
  2. If auth hint exists for this tool, run timeout 10 <tool> auth status 2>&1 (or use the Bash tool's timeout parameter set to 10000ms) and check exit code. If exit code != 0, try timeout 10 <tool> auth check 2>&1. If both fail, mark auth as "FAIL". If the command times out, mark auth as "TIMEOUT". If no auth hint, mark as "—"

MCP servers:

  1. Use ToolSearch with query +<server> to check if any mcp__<server>__* tools are available in the session
  2. If tools found → installed = "yes". If auth tool was referenced in the skill (e.g. auth_test), call it to verify. Otherwise auth = "—"
  3. If no tools found → installed = "no", auth = "—"

gh extensions:

  1. Run gh extension list once (cache the result)
  2. Check if each extension name appears as a suffix in any line of the output (e.g., the line contains /<extension-name> or ends with the extension name)

Phase 4: Present Results

Print a table grouped by skill:

## Prerequisite Check

| # | Skill | Tool | Type | Installed | Auth |
|---|-------|------|------|-----------|------|
| 1 | slack-triage | slack | mcp | yes | yes |
| 2 | meeting-scheduler | gws | cli | yes | FAIL |
| 3 | daily-report | gh | cli | yes | yes |
| 4 | daily-report | slackcli | cli | no | — |
| 5 | oss-release | oss-watch | gh-ext | no | — |

Summary: N skills checked, M tools found, X issues
  • "yes" = check passed
  • "FAIL" = check ran and failed
  • "—" = not applicable (not installed, or no auth pattern detected)
  • Number each row for reference in the fix phase

If no issues found, print "All prerequisites satisfied." and stop.

Phase 5: Interactive Fix

If issues exist, ask: "Want me to fix any of these? (e.g. '2, 4' or 'all' or 'done')"

For each selected item, attempt the appropriate fix:

TypeIssueFix Strategy
clinot installedTry brew install <tool>. If brew doesn't have it, search web for install instructions
cliauth FAILRun <tool> auth login (interactive)
mcpnot availableTell user: "Add the <server> MCP server to your Claude config. I can't auto-configure this."
gh-extnot installedRun gh extension install <owner/repo> — search GitHub for the extension first

After each fix attempt, re-run the check for that item and show the updated row.

When all selected fixes are attempted, re-print the full table with updated results.

Error Handling

  • If ~/.claude/skills/ doesn't exist, report "No skills directory found at ~/.claude/skills/" and check only .claude/skills/
  • If a tool's auth check hangs (>10s), kill it and mark as "TIMEOUT"
  • If brew is not installed, suggest manual install instructions instead

Examples

Example 1: Check all skills

User: "check prerequisites"
Action: Scan all skills, parse deps, check each, show table

Example 2: Check specific skills

User: "/skill-prereq-check slack-triage meeting-scheduler"
Action: Scan only those two skills, parse deps, check, show table

Example 3: Fix issues

User: (after seeing table) "fix 2, 5"
Action: Run `gws auth login` for item 2, `gh extension install` for item 5, re-check

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.