Authentication and token management service with admin UI for managing bearer tokens with host-based access control
Wicket is an authentication token management service, not a general-purpose MCP server. Tool definitions lack critical structure: no input schemas visible in source (auth, healthz, readyz, index have empty/missing schemas); descriptions are present but generic (10-100 chars, below 194-char baseline); no output schemas documented; no error handling guidance; no parameter type definitions. The 'auth' tool accepts authorization headers but schema shows only string types without constraints. 'create_token' has basic descriptions but no validation rules, ranges, or enum constraints. No tool demonstrates idempotent operation patterns or recovery guidance. This server appears designed as a FastAPI service, not as an MCP tool provider with LLM-optimized definitions.
Validates bearer token against allowed hosts. Accepts GET, POST, PUT, DELETE, PATCH, HEAD, OPTIONS requests. Returns 'OK' on successful validation or 401 on invalid token.
Creates a new bearer token with specified hosts, optional email, comment, and TTL. Returns the generated token.
Deletes a token by its hash. Returns 204 No Content on success.
Health check endpoint that returns 'ok' status
Admin UI index page. Returns HTML response with token management interface.
Lists all active tokens with pagination support. Returns tokens with metadata including hosts, email, comment, creation time, and TTL.
No input schemas visible for 4 of 7 tools (healthz, readyz, auth, index).
Parameter descriptions lack type constraints, ranges, and validation rules. 'ttl_seconds' says 'must be >= 0' but no enum/pattern. 'page' and 'size' lack min/max bounds in schema.
No output schemas documented. LLMs cannot infer what fields create_token, list_tokens, or auth return. Breaks downstream tool chaining.
Tool descriptions are generic and under 50 chars for healthz/readyz/index. Descriptions must be 10 - 1024 chars with WHAT/WHEN/WHY context for LLM selection.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 41 | <=2025-11-25 | v2 |
Readiness check endpoint that verifies storage backend health. Returns 'ok' if store is healthy, 503 if unavailable.
No error handling guidance. delete_token returns '204 No Content' but no recovery path if token_hash is invalid. No actionable error messages for LLM.