codetopo — code graph for AI agents

codetopo

Registry identity: mcp-name: io.github.nrupala/codetopo

See your codebase the way an agent sees it. codetopo turns a source repository into a queryable code graph — symbols, calls, references, and structure — with JSON answers an AI coding agent can act on directly.

[OpenSSF Best Practices] [OpenSSF Scorecard]

codetopo holds the OpenSSF Best Practices Passing badge — 100% of 67 criteria.

You are here because an agent just touched your codebase and you want to answer structural questions fast: what calls this function? what breaks if I change this? how are these two modules connected? codetopo indexes the tree once and answers those questions as data — counts first, full lists only when asked — over a CLI, an HTTP API, or the MCP tools your agent already speaks.

It is deliberately domain-blind: it reports measurements — counts and rankings of structural facts, each with its method stated — and it never scores code quality or holds opinions about what the code means. It answers structural questions about source trees, and only that. Domain meaning belongs to the agents and teams composing on top.

codetopo is a product of AIMLDS (2351485 ALBERTA INC.) — the house behind Conduit, managed MCP connector rails where every call is routed, metered, and receipted.

[Illustrative mock-up of a codetopo dependency graph (not actual tool output)]

Live demo: codetopo.aimlds.org — codetopo's real map of this repository, drawn by codetopo itself and refreshed on every push to main.

60-second quickstart

Requires a stable Rust toolchain (install via rustup).

# Install the four binaries: codetopo (CLI), codetopo-server (HTTP API),# codetopo-mcp (stdio MCP server), codetopo-acp (stdio ACP agent).# The packages must be named — a bare install URL fails because the# repository contains multiple binary packages.cargo install --git https://github.com/nrupala/codetopo codetopo-cli codetopo-server codetopo-mcp codetopo-acp
# From crates.io, install the same packages by name — `cargo install# codetopo-cli` for the CLI, `cargo install codetopo-mcp` for the MCP# server. There is no crate named `codetopo`.
# Index a repository (Rust, TypeScript/JavaScript, and Python are parsed via tree-sitter)codetopo index ~/my-project --db graph.db
# Find a symbol's id by name (every surface has this discovery step)codetopo search connect --db graph.db
# Ask what calls what (the package prefix is your directory's name,# exactly as indexed — hyphen included)codetopo descendants my-project::net::Client::connect --db graph.db --expand

That's the whole loop: index once, query as data. Aggregate-first output — totals before lists — keeps token budgets and pipes sane by default; --expand brings the full list when you want it.

Example queries

Against the fixture repository used in the project's own end-to-end verification (docs/phase-2-verification.md) — output shown is the real output of the CLI:

$ codetopo index crates/codetopo-server/tests/fixture-repo --db graph.dbfiles indexed: 3files failed: 0symbols: 6nodes: 15edges: 23diagnostics: 2
$ codetopo stats --db graph.dbnodes: 15  package: 1  module: 3  file: 3  function: 6  external: 2edges: 23  calls: 5  contains: 12  reads: 6
$ codetopo descendants fixture-repo::alpha::entry --db graph.db --expand2 descendants of fixture-repo::alpha::entryfixture-repo::alpha::helper_afixture-repo::alpha::helper_b

A few more things you can ask (same engine, same data):

codetopo ancestors <node-id> --db graph.db     # what depends on this nodecodetopo blast-radius <node-id> --db graph.db  # what a failure here would take with itcodetopo path <from-id> <to-id> --db graph.db  # shortest typed path between two nodescodetopo verify --db graph.db                  # verify the hash-chained audit logcodetopo snapshot --db graph.db --out g.json   # portable frozen JSON export of the whole graphcodetopo restore --snapshot g.json --db g2.db  # restore a snapshot into a fresh database

The path answer renders as typed steps, e.g. a --calls--> x --imports--> b (illustrative; exact ids depend on your index).

Giving it to your agent

HTTP API (codetopo-server)

The same queries over JSON, behind Bearer API-key auth, with per-request metering — one transport for any client, any language:

KEY="demo-key:$(openssl rand -hex 16)"   # a fresh random secret, generated exactly as docker-compose does — never a published example valueecho "$KEY" > api_keys && chmod 600 api_keyscodetopo-server   # listens on 127.0.0.1:8080 by default
curl -H "Authorization: Bearer $KEY" -X POST localhost:8080/v1/index \  -H 'Content-Type: application/json' -d '{"repo_path": "~/my-project"}'# {"index_id": "<uuid>", "files_indexed": 3, "files_failed": 0, "nodes": 15, "edges": 23, "diagnostics": 2}
curl -H "Authorization: Bearer $KEY" \  "localhost:8080/v1/graphs/<uuid>/descendants?node=<node-id>&expand=true"

Twenty endpoints under /v1, one index operation plus nineteen read-only queries: POST /v1/index, GET /v1/graphs/{id}/{descendants, ancestors, blast-radius, path, stats, verify, snapshot, neighborhood, symbol, search, diagnostics, rank, diff, metrics, history, communities, topo-sort, render}, GET /v1/schema. Full contract in schemas/openapi.yaml (OpenAPI 3.1.0), env knobs (CODETOPO_BIND, CODETOPO_DATA_DIR, CODETOPO_KEYS_FILE, CODETOPO_METERING_LOG) documented in schemas/README.md.

MCP (codetopo-mcp)

A stdio JSON-RPC server exposing the same handler layer as twenty tools — the sixteen query/integrity tools (codetopo_index, codetopo_descendants, codetopo_ancestors, codetopo_blast_radius, codetopo_path, codetopo_stats, codetopo_verify, codetopo_snapshot, codetopo_diagnostics, codetopo_rank, codetopo_diff, codetopo_metrics, codetopo_history, codetopo_communities, codetopo_topo_sort, codetopo_render) plus discovery (codetopo_search_symbols, codetopo_get_symbol, codetopo_neighborhood, codetopo_describe_schema) — so an in-process agent learns one surface and gets both. MCP answers are identical to the HTTP answers by construction: both call the same codetopo-core queries.

{ "mcpServers": { "codetopo": { "command": "codetopo-mcp" } } }

ACP (codetopo-acp)

A stdio agent speaking the Agent Client Protocol — for editors and clients that converse with agents instead of calling tools. Open a session on a working directory and ask code-structure questions in plain language:

→ {"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}← {"jsonrpc":"2.0","id":1,"result":{"agent":"codetopo","protocolVersion":"1.0",    "agentCapabilities":{"read_only":true,"code_structure":true},"agentInfo":{"name":"codetopo"}}}→ {"jsonrpc":"2.0","id":2,"method":"session/new","params":{"cwd":"/path/to/repo"}}→ {"jsonrpc":"2.0","id":3,"method":"session/prompt",    "params":{"sessionId":"<id>","prompt":"blast radius of auth::login"}}← {"jsonrpc":"2.0","id":3,"result":{"answer":"Blast radius of 'auth::login': 14 nodes — …"}}

Deterministic intent routing (descendants / ancestors / blast-radius / path / stats / snapshot / verify) straight to the same codetopo-core queries — no LLM, no network, no file writes. Edit requests get an honest read-only refusal. stdout carries only protocol frames; diagnostics go to stderr.

Hosted API (codetopo Hosted)

The same HTTP API, run by us, always on. We host the index; you send a repository (public git URL or archive upload) and query the graph from anywhere with your key — no local install, no server to run.

US$19/month per key.

How it works:

1. Subscribe — checkout is handled by Stripe.
2. After payment you land on a claim page. Claim your API key there — it is shown exactly once, so save it when it appears.
3. Call the API at https://codetopo-api.aimlds.org with Authorization: Bearer <key_id>:<secret>, the same contract as the self-hosted server (see schemas/).

Cancel any time from the Stripe receipt link; the key stays active until the end of the paid period and is then revoked.

This is the hosted form of the same product. The local CLI, MCP server, and ACP agent in this repository stay free and open source (AGPL-3.0) and are unchanged — the hosted tier is for when you want the graph served, not installed.

Architecture

        ┌─────────────┐   tree-sitter    ┌──────────────┐        │ source tree │ ───────────────► │ codetopo-    │   extract symbols,        └─────────────┘   Rust, TS/JS, Py │ extract      │   calls, references                                          └──────┬───────┘                                                 ▼                                          ┌──────────────┐                                          │ codetopo-core│   graph model:                                          │              │   nodes, edges,                                          │              │   traversal queries                                          └──────┬───────┘                                                 ▼                                          ┌──────────────┐                                          │ codetopo-    │   SQLite store +                                          │ store        │   hash-chained                                          │              │   audit log                                          └──────┬───────┘                                                 │              ┌──────────────────────────────────┼─────────────────────────────┐              ▼                                  ▼                             ▼                              ▼       ┌──────────────┐                  ┌──────────────┐               ┌──────────────┐              ┌──────────────┐       │ codetopo     │                  │ codetopo-    │ Bearer auth  │ codetopo-    │ stdio        │ codetopo-    │ stdio       │ (CLI)        │                  │ server       │ JSONL        │ mcp          │ JSON-RPC     │ acp          │ ACP       └──────────────┘                  └──────────────┘  metering    └──────────────┘              └──────────────┘

The CLI, HTTP, MCP, and ACP surfaces are all thin adapters over the same handlers — they cannot disagree. schemas/ keeps the machine-readable contracts (OpenAPI 3.1, MCP tool catalogue) checked against the running servers. The store's audit log follows the trust-primitive pattern used across this product line: a hash-chained, append-only record, verified by codetopo verify / GET …/verify.

Docs

• Development log — how codetopo got here and where it is going, in plain words (kept with the releases; also linked from the demo site)
• docs/render.md — deterministic render: the frozen layout spec (JSON) + SVG rules behind codetopo render
• docs/protocol-proof.md — MCP/ACP conformance evidence: strict stdio clients + official-schema validation
• docs/scale-gate.md — the 10,000-file scale gate and its measured record
• Dockerfile + docker-compose.yml — self-hosted container: the HTTP API over your own indexes
• docs/phase-2-design.md — API design: endpoints, auth, metering, aggregate-first discipline
• docs/phase-2-verification.md — end-to-end verification report (fixture numbers above come from here)
• docs/phase-3-acp-design.md — ACP agent design: wire protocol, intent routing, read-only guarantee
• docs/phase-3-acp-verification.md — ACP verification report
• docs/quickstart-verification.md — clean-machine verification of the quickstart and example queries above
• schemas/README.md — the two agent-facing contracts and where the spec departs from the design doc
• CODE_SCHEMA.md — the graph schema (§6 sets the aggregate-first rule)

Privacy Policy

codetopo runs entirely on your machine. The MCP server (codetopo-mcp), the CLI, and the index databases they create never send your code, your queries, or any usage data anywhere: there are no accounts, no API keys, no telemetry, and no network calls at runtime. Indexing reads only the repositories you point it at and writes only the database paths you give it. To delete your data, delete the index database files (the .db files, or the HTTP server's data directory) — nothing is held on any server, so there is nothing elsewhere to erase. To take your data with you, export it with the snapshot command, which writes the whole graph as one portable JSON file. Questions about privacy: [email protected].

License

• Engine (codetopo-* crates): AGPL-3.0-or-later (see LICENSES/)
• Agent-facing contracts (schemas/: OpenAPI 3.1 spec, MCP tool catalogue): Apache-2.0 (see schemas/LICENSE-APACHE)

Owned by Nrupal Akolkar · Built with Muse by Meta — join Muse, then redeem code 101EEK in Settings within 48 hours of joining and we both earn tokens.