For AI agents 32 MCP tools Open standard

If you’re an agent reading this: here’s the toolbox you always wanted.

codeindexer exposes 32 tools over the Model Context Protocol: semantic + symbolic + exact search in one cascade, a real cross-language call graph (callers, callees, chains, cycles, blast radius), persistent memory cards and solutions, strategic roadmaps, and an on-demand security audit. One local server, every MCP client, no code leaving the machine. This page is the deep technical surface — the spec, not the pitch.

Read the docs Browse the 32 tools

Connect in one command

Claude Code shown below; codeindexer help integration prints the same config for Cursor, Copilot, Windsurf, Codex CLI and Gemini.

bash

claude mcp add --transport http codeindexer \
  --scope user http://127.0.0.1:8978/mcp

Capabilities

One server. 32 tools.
Every workflow.

Your AI agent doesn't see a “code indexer” — it sees a toolbox. Here's what's in it.

Search & navigation

Find symbols, files and patterns; walk the call graph in any direction

group 01 / 05

search_code semantic, symbol, exact, or cascade (all three) — cascade is Pro
read_chunk fetch the indexed chunk that owns a given line
read_file_range precise file slice (cap 500 lines) without leaving MCP
find_references every place a symbol is used — Pro
find_callers walk the call graph backwards (transitive option) — Pro
find_callees walk the call graph forwards (transitive option) — Pro
find_call_chain shortest path between two functions — Pro
find_call_cycles detect circular call paths — Pro
shortest_call_path fastest known path between two symbols — Pro
find_by_signature by parameter types, return type, async, modifier — Pro

Code understanding

Reason about structure, impact, entry points and quality

group 02 / 05

find_dead_code functions nothing calls anymore — Pro
find_complex_functions high cyclomatic complexity offenders — Pro
find_bridges modules connecting distinct subsystems — Pro
find_execution_flows labeled entry-point traces — Pro
rank_functions importance score across the call graph — Pro
review_diff blast-radius analysis of a git diff — Pro
analyze_corpus TF-IDF topic clustering for docs and corpora — Pro
find_related files that co-occur in search results — Pro
find_tropes recurring patterns and idioms across the codebase — Pro
consistency_check cross-file naming, structure and convention checks — Pro
find_citations where docs / READMEs cite a symbol or file — Pro

See full tool reference in docs

Planning, built in

From flailing to a tracked plan.

Plan, review, track, ship — without leaving the code. A full planning surface lives right beside it: cards for todos, decisions, bugs and gotchas, PR-style review threads before a line ships, and roadmaps with checkable success criteria. The first time your agent finds this over MCP, something clicks — here is the workspace it always wanted.

01 / 05

Task & memory cards

Every plan, decision and to-do as a first-class card.

Capture work as cards — todo, decision, bug, gotcha, convention. Each one is linked into the same code graph your agent already searches, so it stays queryable across sessions and never gets lost between chats.

todo decision bug gotcha linked

02 / 05

Review threads on cards

PR-style review — but for the plan, before any code ships.

Open a multi-agent review thread right inside a card. Agents discuss, critique and sign off on the approach first, so the work is vetted before a single line is written.

card chat multi-agent sign-off

03 / 05

Roadmaps with phases

Group cards into phases with real goals.

Bundle cards into phases with concrete goals and success criteria. Watch the done / total counter tick up as an initiative moves — strategy your agent can actually read and act on.

phases goals progress

04 / 05

Checkable success criteria

Progress you can tick, not vibes.

Every success criterion is a checkbox. You — or your agent — tick them off as the work lands, and the roadmap tracks genuine completion instead of optimistic guesses.

checklists criteria tracked

05 / 05

Import & export

Your whole planning vault is portable.

Move cards and roadmaps in and out as portable files. Back them up, version them in git, share them with a teammate, or carry your plan to another machine — nothing is locked in.

import export git-friendly new

>_ Scales from a one-line todo to a multi-phase roadmap — linked into the same graph your agent already searches. Free within limits (8 phases / 42 cards / 64 reports per project), not a paid seat — yours, on your machine.

Languages & formats

Speaks your stack.

Multilingual semantic search out of the box — your queries work in English, Russian, Chinese, German, French, and more (100+ languages).

Full call-graph parsing · 27

Rust Python JavaScript TypeScript PHP Go Java C C++ C# Ruby Swift Solidity Kotlin Scala Dart Lua Bash Julia Haskell Elixir Zig Vue Svelte GLSL HLSL WGSL COBOL*

Symbols plus resolved call / import / type edges. Vue and Svelte through their script block; Ansible playbooks and roles earn the same graph per file. * COBOL ships behind the cobol opt-in build feature.

Symbols & search · 5

YAML Markdown (headings) GraphQL OCaml Verilog / SystemVerilog

Parsed into named symbols — no call-graph extraction.

Recognized & searchable · 36

JSON / JSONL HTML CSS / SCSS / Sass / Less ERB / EJS Jinja SQL TOML Regex Vyper Move Cairo Perl R Erlang Clojure F# Objective-C++ Protobuf Groovy / Gradle Astro Nim Crystal Elm Gleam PureScript Haxe Odin Fortran Pascal Ada Lisp Emacs Lisp Racket Scheme GDScript QuakeC

Line-based chunking for full-text + semantic search — no symbols, no call graph.

Document formats · 7

PDF DOCX XLSX (XLS · XLSM · XLSB · ODS) LaTeX (incl. arXiv .tar.gz) Markdown / docs Telegram exports (JSON) Archives: ZIP · TAR.GZ/BZ2/XZ/ZST · RAR

Integrations

Works with the AI clients
you already use.

If your client speaks MCP, it speaks codeindexer. Battle-tested with:

Client Transport Notes

Cursor HTTP / SSE One-line config, auto-discovery

Claude Code HTTP / SSE · stdio Both claude CLI and project-level .mcp.json

Claude Desktop stdio Zero-config

VS Code Copilot HTTP MCP Discovery toggle in settings

Windsurf stdio Native mcp_config.json support

OpenAI Codex CLI stdio config.toml entry

Gemini / Antigravity HTTP Streamable HTTP endpoint

Also works with

Continue · Cline · OpenCode · Crush · Kimi Code · or any other MCP-capable client (HTTP, SSE, or stdio).

One-command helper

We ship a one-command helper to generate the right config for any of them:

bash

codeindexer help integration

Pick your client — get copy-pasteable JSON, TOML or stdio config in your clipboard.

Security & privacy

Your code stays your code.

127.0.0.1

Local execution

codeindexer runs entirely on your machine. Indexing, embedding, search, audit — everything happens on localhost. The server binds to 127.0.0.1 by default. No source code, no file contents, and no search queries are sent to our servers, ever.

1 req / day

What we do see

Pro and Enterprise licenses make one HTTPS call per day to verify the license. That request contains a license ID and an anonymous machine fingerprint — nothing else. The exact payload is auditable in our open documentation. If the license server is unreachable, the binary continues normally for a 14-day grace window.

0600

Hardening

HTTP transport supports bearer-token authentication, configurable CORS allow-list, and same-origin protection out of the box. Stdio transport is process-level isolated. Index data lives in your user data directory with restricted permissions. License files are stored with 0600 mode on Unix.

GDPR-compatible no personal data leaves your machine

EU AI Act ready fully transparent, auditable, local processing

Optional audit logs append-only log of every tool invocation, per project

No training on your code we don't have it; we can't train on it

Open MCP standard no vendor lock-in; swap us out anytime

Wire it in

Connect it to your client in 60 seconds.

HTTP or stdio transport, works with Cursor, Claude Code, Copilot, Windsurf, Codex CLI and Gemini. Ask your agent what tools it has — it'll list everything codeindexer just gave it.

Read the docs See pricing