An MCP server that provides GitHub OAuth authentication and integration, exposing tools for GitHub operations and user-authenticated AI features
This server exhibits significant quality gaps across naming, descriptions, and schema completeness. While tool names are action-oriented (add, userInfoOctokit, generateImage), parameter descriptions are sparse or missing entirely. The 'add' tool has zero parameter descriptions despite having inputs. The 'userInfoOctokit' tool takes no parameters but lacks clarity on what GitHub user info it returns. The 'generateImage' tool has better parameter documentation but still lacks output schema definition. No tools document their return types or structure. Error handling is absent, no guidance for recovery scenarios. Schema validation exists via Zod but is not visible as JSON Schema documentation for LLM consumption.
Add two numbers the way only MCP can
Generate an image using the `flux-1-schnell` model. Works best with 8 steps.
Get user info from GitHub, via Octokit
Parameter descriptions missing entirely for 'add' tool (parameters 'a' and 'b' have null description in schema). LLMs cannot infer what numeric values mean without explicit descriptions.
No output schemas documented for any tool. 'add' returns a number, 'userInfoOctokit' returns GitHub user object structure, 'generateImage' returns image data, but none of these are documented. LLMs cannot plan downstream operations without knowing response shape.
'userInfoOctokit' tool name is unclear. 'userInfoOctokit' does not follow verb_noun pattern (should be 'get_user_info'). The suffix 'Octokit' is implementation detail leaking into the interface, users don't care what library backs the tool. Additionally, what specific user? The authenticated user? A parameter would clarify.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 5 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 41 | - | v1 |
Tool 'add' description is 'Add two numbers the way only MCP can', vague and marketing-focused rather than actionable. Does not explain parameters, constraints, or when to use vs alternatives. Description length is 34 chars, below the 50-char actionable minimum.
No error handling or recovery guidance. If 'generateImage' fails (rate limit, invalid prompt, API down), tool provides no actionable error message or retry strategy. Pattern compliance missing.
Parameters in 'generateImage' lack minimum/maximum validation messaging. While Zod constraints exist (min 4, max 8 for steps), they are not visible to the LLM in the exposed schema, LLMs rely on description text to understand constraints, not JSON Schema.
'generateImage' tool is dynamically registered based on ALLOWED_USERNAMES set (currently empty). This means the tool may or may not be available at runtime depending on authentication context. LLM cannot reliably discover or reason about tool availability, creates a 'ghost tool' problem.