What this skill does
A raw API error response — a stack trace, an HTTP status with a cryptic JSON body, a webhook that silently failed — usually needs translating before you can act on it, and that translation is where a lot of debugging time quietly goes. This skill takes any raw error output and returns a plain-English diagnosis: what actually went wrong, the most likely root cause, and one concrete next step to try first.
How it works
- Error parsing — reads the raw error text (HTTP status + body, stack trace, log line) and identifies the error type and the layer it came from (auth, rate limit, validation, network, server-side).
- Root-cause reasoning — cross-references the specific error against known patterns for common APIs and general HTTP/REST semantics to propose the most likely cause, not just a restatement of the error text.
- Next-step suggestion — ends with one specific, actionable next step (a header to check, a field to validate, a retry-with-backoff pattern to add) rather than a generic "check your API key" catch-all.
Requirements
- Works on Claude AI (any surface) and on OpenClaw (useful inside a debugging or on-call workflow) — a pure reasoning skill with no external tool dependency.
Setup
- Unzip the package into your Skills directory, keeping the folder structure intact.
- Paste the raw error output (status code, response body, and the request that triggered it if you have it) — see
SKILL.mdfor the exact invocation pattern.
What's included
SKILL.md— the skill definition and diagnosis logicexamples/sample-run.md— worked examples across auth, rate-limit, and validation error types
