All 4 tools have trivial definitions with empty input schemas (parameters: {}). Descriptions are present but minimal (10-50 chars, well below the 194-char baseline). No parameter descriptions exist because no parameters are defined. No output schemas documented. Tools lack error handling guidance, validation, or recovery paths. This server reads as a proof-of-concept rather than production-grade tooling.
Get all available categories for Chuck Norris jokes
Get a random Chuck Norris joke
Get a random dad joke
Get a random Yo Mama joke
All tools have empty input schemas (parameters: {}).
Tool descriptions are 10 - 50 characters (e.g., 'Get a random Chuck Norris joke'). Baseline is 194 chars. Descriptions lack WHEN to use, prerequisites, or what the tool returns.
No output schemas documented. LLMs cannot plan downstream calls or extract fields without knowing the response structure. All tools return {content: [{type: 'text', text: ...}]} but this is not declared in the tool definition.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 40 | 2026-07-28+ | v2 |
No error handling or recovery guidance. If an external API (chucknorris.io, icanhazdadjoke.com, yomama-jokes.com) fails, the tool returns a raw error with no actionable next step for the LLM.
No input validation or timeout handling. External API calls lack explicit timeouts; a hung request blocks the agent indefinitely. No validation of fetch responses (e.g., checking response.ok or catching JSON parse errors).