Whole-repo code intelligence for AI agents, served over MCP.
AI coding agents are great at the file they're looking at and blind to the other 2,000. Idencify gives them the whole-repo picture: connect a GitHub repository and it builds a cross-language call graph with semantic search — served through 14 MCP tools to Claude Code, Cursor, and any MCP client.
This is the public documentation repository for the hosted server at
https://mcp.idencify.com/mcp. Sign up at idencify.com — free tier included.
claude mcp add --transport http idencify https://mcp.idencify.com/mcpAuthentication is OAuth 2.1 — your MCP client opens the browser, you approve, done. API-key access (REST) is also available from the dashboard.
Related MCP server: knowing
Tool | What it answers |
| "What breaks if I change this symbol?" — transitive blast radius |
| Callers/callees N levels deep, cross-file and cross-language |
| Definition + dependency context for a symbol |
| Assembled, ranked context pack for a task |
| Hybrid semantic + keyword + identifier search |
| Structural pattern occurrences across the repo |
| Dependency cycles, ranked by severity |
| Architecture boundary violations |
| Repo/module health grade with hotspots |
| Exact graph diff between two commits (adds/removes/breaking) |
| Human-readable changelog from graph changes |
| Query indexed git history |
| File summary with symbols and relations |
| Feedback loop — rate retrieved context to improve ranking |
Rust, TypeScript, JavaScript, Python, Go, Java and C++ — all indexed into one unified graph, with enhanced compiler-level precision for Rust, TypeScript, Python and Go. Re-indexing never loses relations: the graph can only get better over time.
Connect a repository (GitHub App with read-only access, or REST API for public repos).
Idencify indexes it into a whole-repo graph: symbols, calls, imports, containment and implementation relations, plus semantic search — updated automatically on every push.
Your agent queries the graph over MCP — every answer is backed by the indexed graph, not guesses.
Website / signup: idencify.com
Dashboard: app.idencify.com
MCP endpoint: https://mcp.idencify.com/mcp (Streamable HTTP; SSE at /mcp/sse)
Contact: hello@idencify.com
Free ($0) · Plus ($9/mo) · Pro ($29/mo) — see idencify.com for limits.
List symbols and files affected by changing a given symbol (reverse dependencies).
Inputs: no named parameters.
Verify architectural boundaries between modules and layers.
Inputs: no named parameters.
Find dependency cycles between symbols and files.
Inputs: no named parameters.
Walk callers/callees from a symbol up to a given depth.
Inputs: no named parameters.
Locate structural code patterns across the repository.
Inputs: no named parameters.
Assemble task-relevant code context for a natural-language question.
Inputs: no named parameters.
Fetch the full source and metadata of a symbol by name or id.
Inputs: no named parameters.
Diff the code graph between two points in time (added/removed symbols and edges).
Inputs: no named parameters.
Score repository health: dead code, cycles, coupling, coverage of the graph.
Inputs: no named parameters.
Feedback loop: rate the usefulness of previously returned context.
Inputs: no named parameters.
Search past ingestion/sync history of the repository.
Inputs: no named parameters.
Human-readable changelog derived from graph changes between syncs.
Inputs: no named parameters.
Summarize a file's symbols, relations and role in the codebase.
Inputs: no named parameters.
Hybrid semantic + keyword search over all symbols in the indexed repository.
Inputs: no named parameters.
Trigger re-indexing of the connected repository.
Inputs: no named parameters.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 15 tools
Each tool targets a distinct aspect of code analysis—search, definition, context, impact, graph traversal, cycle detection, boundary checks, pattern finding, file summary, health score, graph diff, changelog, history, feedback, and sync. Even seemingly related tools like analyze_impact and expand_call_graph are clearly separated by their focus on reverse dependencies versus call graph walking.
The majority of tools follow a verb_noun pattern (get_definition, analyze_impact, detect_cycles), but several deviate—symbol_search, health_score, graph_diff, and semantic_changelog use reversed or noun-first conventions, weakening predictability across the set.
With 15 tools, the server is at the upper edge of the ideal range, yet each tool serves a distinct and necessary function for a comprehensive code intelligence platform, ensuring no tool feels redundant or extraneous.
The tool set covers a full lifecycle: ingesting/triggering syncs, searching and retrieving symbols, analyzing dependencies and cycles, enforcing architecture, finding patterns, summarizing files, assessing health, tracking changes over time, and even incorporating feedback—leaving no obvious gaps for its domain.
Code intelligence platform for AI agents. 20 tools for architecture, security & impact analysis.
Hosted code graph over MCP: exact callers, dependencies, and cross-repo blast radius for AI agents.
AI-powered codebase analysis — call graphs, security, dead code, complexity. 150+ tools.
Shared context for your codebase: a hosted code knowledge graph over MCP. Free plan, OAuth sign-up.
