Fast code intelligence engine for AI coding agents — single static binary MCP server
Critical gaps across all dimensions. Tool descriptions are present but minimal (10-50 chars, well below the 194-char baseline). Input schemas visible but lack parameter descriptions, most params have only type and name, no guidance on format, constraints, or valid values. No output schemas documented. Error handling absent. Destructive tool (delete_project) has no confirmation pattern. Parameter naming inconsistent: 'query' vs 'pattern' for search tools; 'project' appears in 9/10 tools but lacks description in most. No tool annotations (readOnlyHint, destructiveHint). Composition issues: search_graph, query_graph, and search_code overlap functionally without clear differentiation. STDIO transport caps protocol readiness at 50, further limiting production viability.
Delete an indexed project
Get the high-level architecture of a project
Get a code snippet for a symbol or file location
Get the schema of the code graph
Index a repository to build a code graph
List all indexed projects
Query the code graph with structured queries
Search for code patterns and text in the codebase
Parameter descriptions missing or trivial. 'query', 'project', 'pattern', 'symbol' appear across tools with minimal guidance on format, constraints, or valid values. LLMs cannot infer whether 'query' expects regex, natural language, or structured syntax.
No output schemas documented. Tools return unspecified structures. LLMs cannot plan downstream calls or extract required fields (e.g., does search_graph return file paths, line numbers, confidence scores?).
Destructive tool (delete_project) lacks confirmation or dry-run pattern. No error guidance. Agents can irreversibly delete indexed projects without safeguards.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | <=2025-11-25 | v2 |
Search the code graph for symbols and definitions
Trace a path through the code graph between two symbols
Overlapping tool semantics. search_graph, query_graph, and search_code all search for code patterns but lack clear differentiation. LLMs will struggle to choose the right tool. 'query_graph' is vague, does it accept SQL, GraphQL, or a custom DSL?
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Clients cannot infer which tools are safe to retry or which modify state. index_repository (WRITE) and delete_project (DESTRUCTIVE) lack explicit risk markers.