Programming-first context engine for Claude Code — TypeScript-powered codebase indexing, dependency graphing, symbol-level search, and zero-token context injection via hooks.
CCS exposes 6 tools with basic descriptions and partial schemas. Tool names follow verb_noun convention (index, update, context, search, graph, init), which is good. However, parameter descriptions are minimal or absent in most tools. Input schemas are present but lack depth, many parameters have no description text. Output schemas are entirely undocumented. Error handling is not visible in the provided code. The server appears to be a custom HTTP implementation without standard MCP registration patterns visible in the source. Tool definitions are inferred from CLI argument parsing rather than explicit MCP schema registration, which caps individual tool scores at 50 per the hard rules.
Query the CCS index to find relevant files and symbols for a given user prompt, returning formatted context for injection into Claude's prompt
Analyze and return the dependency graph of the indexed codebase, showing import relationships and centrality rankings
Index a codebase project, creating a .ccs/index.json file with file nodes, dependency graph, symbol index, and project statistics
Initialize CCS in a project or globally, installing skills, agents, templates, hooks, and MCP configuration
Search the CCS index for files and symbols matching a query pattern
Incrementally update the CCS index for a single modified or deleted file
Output schemas completely undocumented. No tool returns a documented schema, LLMs cannot plan downstream calls or extract structured data. Critical for agent composition.
Parameter descriptions missing or trivial. 'root' appears in 5 tools with identical minimal description. 'incremental' has no description. LLMs cannot infer parameter semantics from names alone.
Tool definitions inferred from CLI parsing, not explicit MCP schema registration. No visible MCP server implementation, only .mcp.json config pointing to external HTTP endpoint. Cannot verify actual schema compliance.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | 2026-07-28+ | v2 |
No error handling guidance visible. Tools perform file I/O and indexing but provide no recovery hints. E.g., 'index' could fail if root is invalid, but no error message tells the LLM what to do next.
No pagination or result limits documented. 'search' and 'graph' could return large result sets. No limit, offset, or cursor parameters visible. Large responses risk context window exhaustion.