SituationalMITOriginal

AI Endpoint Smoke Test

Fires a smoke test at AI-powered API endpoints (Claude, GPT, custom agents) on live sites — flags 5xx, timeouts, broken auth.

Entry verified April 20, 2026

The short answer

If your sites have AI features (Claude-powered tools, prompt optimisers), those endpoints break silently until a user complains. This skill tests them proactively.

When to use it

Daily or weekly scheduled run. After a provider pushes an SDK update.

Setup

  1. 1

    Save as ~/.claude/commands/ai-smoke-test.md.

  2. 2

    Configure: list of endpoints + expected response shape.

  3. 3

    Invoke: /ai-smoke-test.

Example

You: /ai-smoke-test
Claude: 3 endpoints tested. ✓ Vendors.ie /api/explain. ✓ FluentAtDesk /api/rewrite. 🔴 ReadingCurator /api/recommend — 500, Anthropic API key expired.

Source & attribution

Author
Bryan Collins
Licence
MIT
Type
Original

Original pattern published under MIT — attribution preserved by convention, not licence requirement.

Caveats

Don't include sensitive test inputs in the skill — inject from a separate secrets file.

Related skills

Security Audit

Sweeps a repo for OWASP-top-10 issues, hardcoded secrets, insecure dependencies, and auth logic gaps.

Dependency Audit

Checks every dependency's licence — flags GPL contamination, commercial-incompatible licences, missing attribution.

Browse more skills

47 skills across 10 categories, all licence-verified.

Back to catalogueMore in QA, review, security
Read the guide