Standalone MCP server for LLM Wiki — 15 tools working directly against a wiki directory
Single tool 'discover_entities' has a reasonable description (118 chars, within 10-1024 range) but critical schema deficiencies. Input schema is present with two parameters (wiki_root, entity_type), both typed as strings with descriptions. However, no output schema is documented, LLMs cannot infer what fields the response contains or how to chain downstream calls. The description mentions 'Uses the Python sidecar for zero-subprocess execution' and 'Builds entity registry from wiki pages via link_suggest.py's build_entity_registry pipeline', implementation details that don't explain WHAT the tool returns or WHEN to use it. No error handling guidance. No pagination support despite likely returning lists. Tool naming 'discover_entities' is acceptable (verb_noun pattern) but generic, 'list_wiki_entities' or 'search_wiki_entities' would be clearer. The server claims 15 tools in package.json description ('15 tools working directly against a wiki directory') but only one tool is visible in the provided source code (discover-tool.ts). This suggests either incomplete source submission or tool definitions are not visible for evaluation.
Discover all entities registered in the wiki. Uses the Python sidecar for zero-subprocess execution. Builds entity registry from wiki pages via link_suggest.py's build_entity_registry pipeline.
No output schema documented. LLMs cannot determine what fields the response contains (entity names, types, metadata, counts, etc.) or how to chain results to downstream tools.
Description lacks WHAT the tool returns and WHEN to use it. Mentions implementation details (Python sidecar, link_suggest.py) instead of user-facing behavior. Does not state if results are paginated, how many entities are returned, or what fields are included.
No pagination support visible. If wiki contains hundreds of entities, returning all at once will exhaust context window. Missing limit, offset, or cursor parameters.
Only 1 tool visible in source code, but package.json claims 15 tools. Either source submission is incomplete or tool definitions are not explicitly registered in the provided code.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 49 | 2026-07-28+ | v2 |
No error handling guidance. If wiki_root is invalid or entity_type filter matches nothing, what does the tool return? How should the LLM recover?