agentleFS
notquenotque/vexjoy-agent/skills/infrastructure/endpoint-validator/SKILL.md

endpoint-validator

Skill448 starsChanged 4 days agoMITView on GitHub

Deterministic API endpoint validation with pass/fail reporting.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 3 are distinctive, and 6 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.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read repository CLAUDE.md for base URL conventions or env var names.

    line 46
  2. Very specific:

    Search for endpoint config in order: endpoints.json in project root, tests/endpoints.json, inline specification from user.

    line 48
  3. Very specific:

    Use hostnames or ${ENV_VAR} in base_url, not hardcoded IPs.

    line 71
  4. Distinctive:

    Execute, do not reason: Run the request. Paste the exit code and output.

    line 78
  5. Distinctive:

    Test endpoints sequentially.

    line 80
  6. Very specific:

    Check status code against expect_status.

    line 82
  7. Very specific:

    If expect_key set, parse JSON and check key exists.

    line 83
  8. Very specific:

    If max_time set and elapsed exceeds it -> SLOW.

    line 84
  9. Very specific:

    Gate: all endpoints tested, each has PASS/FAIL/SLOW verdict.

    line 89

≈ “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

Endpoint Validator Skill

Deterministic HTTP endpoint validation: discover config, test each endpoint against expectations, report pass/fail with CI-compatible exit codes.

Deep References

Load on demand when the signal matches.

SignalReferenceContent
Security header WARNs, HSTS/CSP/X-Framereferences/security-headers.mdHeader checks, required values, remediation
Config errors, hardcoded IPs, timeoutsreferences/endpoint-config-preferred-patterns.mdConfiguration failure modes and fixes
401/403 failures, auth patternsreferences/auth-endpoint-patterns.mdBearer, API-key, cookie auth validation

Instructions

Phase 1: DISCOVER

Read repository CLAUDE.md for base URL conventions or env var names.

Search for endpoint config in order: endpoints.json in project root, tests/endpoints.json, inline specification from user. Prefer version-controlled config over ad-hoc lists.

Config shape:

{
  "base_url": "http://localhost:8000",
  "endpoints": [
    {"path": "/health", "expect_status": 200},
    {"path": "/api/v1/users", "expect_key": "data", "timeout": 10},
    {"path": "/api/v1/search?q=test", "max_time": 2.0}
  ]
}

Endpoint fields: path (required), expect_status (default 200), expect_key (top-level JSON key), timeout (default 5s), max_time (SLOW threshold), method (default GET), headers (per-endpoint).

Rules:

  • Warn before POST/PUT/DELETE against production base URLs.
  • Use hostnames or ${ENV_VAR} in base_url, not hardcoded IPs.
  • Confirm base URL reachable before the full suite.

Gate: config parsed, base URL reachable, at least one endpoint defined.

Phase 2: VALIDATE

Execute, do not reason. Run the request. Paste the exit code and output.

Test endpoints sequentially. For each:

  1. Send request with configured method and timeout.
  2. Check status code against expect_status. Mismatch -> FAIL.
  3. If expect_key set, parse JSON and check key exists. Missing/invalid -> FAIL.
  4. If max_time set and elapsed exceeds it -> SLOW.
  5. On non-localhost URLs, check security headers (HSTS, CSP, X-Content-Type-Options, X-Frame-Options). Missing -> WARN (not FAIL). Skip HSTS on HTTP-only base URLs. Skip X-Frame-Options if CSP has frame-ancestors.

Failure handling: connection refused -> FAIL "Connection refused"; timeout -> FAIL "Timeout after Ns"; invalid JSON on expect_key -> FAIL "Invalid JSON response".

Gate: all endpoints tested, each has PASS/FAIL/SLOW verdict.

Phase 3: REPORT

Format:

ENDPOINT VALIDATION REPORT
==========================
Base URL: http://localhost:8000
  /api/health                    200 OK      45ms
  /api/products                  500 FAIL   "Internal Server Error"
  /api/slow                      200 SLOW   3.2s > 2.0s threshold

SUMMARY: Passed 13/15 (86.7%), Failed 1, Slow 1

Exit 0 if all passed. Exit 1 if any failed.

Error Handling

ErrorCauseFix
Base URL unreachableWrong port or service downss -tlnp to confirm port
All endpoints timeoutWrong host or proxy issuecurl -v a single endpoint
JSON parse failure on expect_keyNon-JSON response (HTML/XML)Remove expect_key or check Content-Type
FAIL on intentional 404Default expect_status is 200Set "expect_status": 404
401 on auth endpointMissing/expired credentialsAdd auth header; see references/auth-endpoint-patterns.md

CI Integration

# Pre-deployment gate with curl
jq -r '.endpoints[].path' endpoints.json | while read path; do
  curl -sf "http://localhost:8000$path" > /dev/null || { echo "FAIL: $path"; exit 1; }
done

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.