Rust/PostgreSQL/ParadeDB code intelligence system that builds a semantic knowledge graph from any codebase, providing MCP tools for code search, analysis, and context retrieval
Cartograph exposes 13 tools with basic schemas and descriptions, but falls short of production quality. Most tools have descriptions (10/13), but they are generic and lack actionable context for LLM selection. Schemas are present but minimal, most parameters lack type constraints, enums, or validation guidance. No tool descriptions explain WHEN to use them vs. similar tools (e.g., cartograph_search vs. find vs. cartograph_context). Error handling is absent, no recovery guidance, no categorization of retryable vs. fatal errors. Parameter descriptions are sparse (e.g., 'query' in cartograph_search lacks format/length guidance). No output schemas documented. Tool naming is inconsistent: verb_noun (cartograph_status, cartograph_search) mixed with bare nouns (files, node, graph). Composition is weak, tools like cartograph_review and cartograph_admin lack clear boundaries. Security: cartograph_admin accepts unbounded 'operation' strings with no enum constraint, inviting injection attacks.
Find affected symbols and files from code changes
Retrieve symbols and references at a specific source range
Bounded lifecycle work for index management, database operations, and maintenance
Retrieve semantic context for a coding task with evidence assembly and deterministic retrieval
Full workflow and tool-routing guide for complete coding tasks
Analyze code changes against the indexed codebase and provide review feedback
Find exact, lexical, or semantic evidence in the current generation
Report current generation counts and live-source freshness
cartograph_admin accepts unbounded 'operation' string with no enum constraint or validation. Invites injection attacks and makes LLM selection ambiguous.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., does cartograph_search return file_id, line_number, confidence?).
Overlapping tool names and unclear boundaries: cartograph_search, find, and cartograph_context all appear to search/retrieve code. Descriptions do not explain when to use each instead of the others.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 44 | 2025-06-18+ | v2 |
Find entry points and hotspots in the codebase
Query and retrieve file information from the indexed codebase
Find exact, lexical, or hybrid evidence in the current generation
Query graph structure and relationships
Query graph nodes and their properties
Parameter descriptions lack actionable constraints. 'query' in cartograph_search, find, and cartograph_context lacks format, length, or example guidance. LLMs cannot validate input before calling.
No error handling or recovery guidance. Tools do not indicate which errors are retryable, which require user input, or which are fatal. LLMs cannot self-correct on failure.