Codebase context and code search tools for coding agents
Context layers and code search that help agents understand large codebases with fewer tokens. Of the 964 tools on our writing code page, 41 fit this narrower job. By pricing: 32 open source, 2 source-available, 3 free or freemium, 2 paid, usage-based or enterprise, 2 pricing on the maker's site. The most-starred open repos here are Claude Code (153,919 stars), Graphify (139,123 stars), Archify (94,965 stars). Most come from the coding (26), MCP servers (8), agent infra (2) categories.
Updated
Claude Code
Claude Code is Anthropic's agentic coding tool that lives in your terminal, understands your codebase and carries out routine tasks, explains complex code and handles git workflows through natural-language commands. It also runs in IDEs and on GitHub when you tag @claude, and installs via a native script, Homebrew or WinGet (the npm install is deprecated).
Added
Graphify
Graphify is an Apache-2.0 /graphify skill for Claude Code, Cursor, Codex and Gemini CLI that maps a whole project, including code, docs, SQL schemas, configs and PDFs, into a knowledge graph the agent can query instead of grepping through files. Code is parsed locally with tree-sitter, so the map is deterministic, needs no LLM or vector store, and nothing leaves your machine, and every edge in the graph says why it exists. A separate paid Graphify Cloud adds always-on team memory with a 14-day trial.
Added
Archify
Archify is an Apache-2.0 agent skill for Cursor, Claude Code, Codex CLI and OpenCode that turns an idea, plan or whole codebase into an interactive HTML diagram you can explore, follow source links through and share, from repository maps to learning paths and travel plans.
Added
Understand-Anything
Turns any codebase, knowledge base or docs folder into an interactive knowledge graph you can explore and question, working inside Claude Code, Codex, Cursor, Copilot and Gemini CLI.
Added
CLI-Anything
CLI-Anything, from the HKU Data Science lab (HKUDS), is an Apache-2.0 project that turns existing desktop software, codebases and web APIs into agent-native command-line tools, so coding agents like Claude Code can drive programs such as Blender, GIMP, LibreOffice or CAD apps through structured commands with previews instead of screenshots. Its CLI-Hub (pip install cli-anything-hub) lets agents browse and install community-built CLIs, and contributors can add new harnesses by pull request.
Added
Aider
Aider is an Apache-2.0 AI pair programmer in your terminal: you start a new project or work in an existing repo, and it edits files across 100+ languages using a map of your whole codebase, then commits each change to git with a sensible message. It works with Claude, DeepSeek, OpenAI models and almost any other LLM, including local models, and also supports voice input, images and web pages as context and linting and testing after each edit.
Added
Open Code Review
Open Code Review is Alibaba's Apache-2.0 AI code review CLI (ocr), used internally by tens of thousands of Alibaba developers before it was open-sourced. It reads Git diffs and sends changed files to an agent that can read whole files, search the codebase and inspect related changes, then posts structured line-level comments; deterministic pipelines and a built-in multi-language ruleset catch issues like null-pointer risks, thread safety, XSS and SQL injection, and ocr scan reviews whole files or folders without a diff. It works with any OpenAI- or Anthropic-compatible model endpoint.
Added
Codebase Memory MCP
MCP server that indexes codebases into a persistent knowledge graph so coding agents can look up structure quickly.
Added
Context Mode
Context window optimizer for coding agents that sandboxes tool output and persists session memory.
Added
code-review-graph
Local code knowledge graph built with Tree-sitter that serves compact review context over MCP, so coding agents read only the files a change touches.
Added
Composio
Composio gives AI agents more than 1,000 pre-authenticated toolkits for apps such as Gmail, Slack, GitHub and Salesforce, with per-user sessions, managed OAuth, tool search, triggers and a sandboxed workbench, so an agent can act in a user's accounts without you building each integration. The MIT-licensed monorepo ships TypeScript and Python SDKs, a CLI for searching and running tools from the shell, and adapters for the OpenAI Agents SDK, Claude Agent SDK, Vercel AI SDK, LangChain and more; the hosted platform has a free tier.
Added
Repomix
CLI and web tool that packs an entire repository into a single AI-friendly file (with an MCP server) so coding agents and LLMs get full codebase context.
Added
Dagger
Automation engine to build, test and ship any codebase. Runs locally, in CI, or directly in the cloud
Added
HumanLayer
Tools for getting coding agents to solve hard problems in complex codebases.
Added
Claude Context
Zilliz MCP plugin that adds semantic code search to Claude Code and other coding agents, using your whole codebase as context.
Added
Trail Code
Context layer that makes Claude Code, Cursor, Codex and other coding agents faster and cheaper on large codebases.
Added
Repowise
Codebase intelligence tool that indexes code, call graph, git history and docs locally and answers questions for you and your coding agent over MCP.
Added
Semble
Fast code search for agents that uses far fewer tokens than grep and file reads.
Added
Potpie
Builds a knowledge graph of your codebase so agents can help across the software lifecycle.
Added
Code-Graph-RAG
Graph-based RAG for monorepos that lets agents query and edit multi-language code.
Added
Fallow
Codebase intelligence for TypeScript and JavaScript in a single Rust binary. It reports health, complexity hotspots, duplication, circular dependencies, architecture boundaries and unused code, and ships as a CLI, a GitHub Action, an LSP, an MCP server and a VS Code extension.
Added
LazyCodex
Agent harness for Codex that adds project memory, planning, execution and verified completion for complex codebases.
Added
SocratiCode
Codebase context engine for AI agents with hybrid semantic search, dependency graphs and impact analysis, available as MCP, plugin or skill.
Added
TrustGraph
TrustGraph is an Apache-2.0 ontology runtime for AI agents: it turns your documents and data into ontology-grounded knowledge graphs built on W3C standards (RDF, OWL, SHACL, PROV-O) and governs what each agent and user can see. It is aimed at enterprises where agent decisions carry real risk, keeps your ontology, knowledge and audit trail portable and self-hosted, and has a playground, Docker images and a PyPI package.
Added
CodeBoarding
Generates interactive architecture diagrams of codebases.
Added
Shippie
Extendable code review and QA agent that reads your diff, explores the codebase and posts focused review comments.
Added
Gemini MCP Tool
MCP server that lets Claude and other assistants call the Gemini CLI to analyze large files and codebases with its big context window.
Added
AOCI-CODE
AOCI-CODE keeps a persistent, Git-versioned map of your whole codebase and database schema that your coding agent writes and a local MCP server governs. Agents in Codex, Claude Code, Cursor or OpenCode read the map once and know what each file is for, what it depends on and what must not break, instead of re-reading the repo on every task. It can index an existing codebase of up to about half a million lines and report how well it knows each area before taking over development.
Added
Webbrain
Webbrain is an open-source agent project that helps builders handle AI browser agent for Chrome and Firefox (monorepo).
Added
Repobrain
๐ง RepoBrain (formerly Antigravity) โ Give your repo a brain. ChatGPT for your codebase: works in Claude Code, Cursor, Codex, Windsurf & more.
Added
Axiomcodegraph
AxiomCode Graph turns a repository into a knowledge graph built with formal methods: for every function it derives exactly who calls it and what it calls, including calls through interfaces and subclasses that grep misses. Coding agents then explore, search and edit from that map to see what each change breaks, saving time and tokens. It runs locally with a CLI.
Added
nanocode
nanocode is a small SDK for embedding frontier-model agents inside your own apps and tools. It provides the agent loop, tool calling and session handling for models such as Claude and GPT in a compact codebase, so developers can ship a capable embedded agent without adopting a heavy framework.
Added
Octocode
Octocode is an MCP server that gives agents a way to handle code indexer with semantic search and knowledge graphs.
Added
Roam Code
Roam Code is an MCP server that gives agents a way to handle local codebase intelligence CLI + MCP server for AI coding agents: SQLite.
Added
Code Context Engine
Code Context Engine is an MCP server agents call to index your codebase.
Added
RoboSystems
RoboSystems is an MCP server that gives agents a way to handle accounting knowledge graphs: SEC XBRL filings, QuickBooks ledgers, reports and forecasts over.
Added
Spiderbrain
Spiderbrain is an MCP server that gives agents a way to handle deterministic context layer for your codebase: change impact, blast radius, answers.
Added
Augment Code
Agentic software development platform for engineering teams, with coding agents that use deep context from large codebases.
Added
cubic
AI code review that catches bugs in complex codebases on every pull request.
Added
Greptile
AI code reviewer that reads your whole codebase to review pull requests and catch bugs before merge.
Added
Ml) Persistent Agent Memory & Code Localization
Sub-millisecond persistent memory, behavioral code search, and bug localization for AI agents.
Added
GitHub repos (38)
| Repo | GitHub stars | Licence | Last push |
|---|---|---|---|
| 153,919 | Proprietary (All rights reserved; Anthropic Commercial Terms of Service) | 2026-10-30 | |
| 139,123 | Apache License 2.0 | 2026-10-27 | |
| 94,965 | Apache License 2.0 | โ | |
| 88,992 | MIT | 2026-10-26 | |
| 57,786 | Apache License 2.0 | 2026-09-22 | |
| 49,825 | Apache License 2.0 | 2026-05-22 | |
| 49,071 | Apache License 2.0 | 2026-10-30 | |
| 47,830 | MIT | 2026-10-29 | |
| 32,905 | Elastic License 2.0 (source-available) | 2026-10-30 | |
| 32,498 | MIT | 2026-10-15 | |
| 30,828 | MIT | 2026-10-29 | |
| 29,080 | MIT | 2026-10-28 | |
| 16,550 | Apache License 2.0 | 2026-10-30 | |
| 12,712 | Apache-2.0 | 2026-10-29 | |
| 12,660 | MIT | 2026-07-14 |
Top 15 of 38 by stars. Stars and licence come from each repo, checked when the listing was last verified.
FAQ
What are the best AI tools for writing code focused on codebase context and code search?
We list 41. The first three by GitHub stars and featured status are Claude Code: Claude Code is Anthropic's agentic coding tool that lives in your terminal, understands your codebase and carries out routine tasks, explains complex code and handles git workflows through natural-language commands. It also runs in IDEs and on GitHub when you tag @claude, and installs via a native script, Homebrew or WinGet (the npm install is deprecated).; Graphify: Graphify is an Apache-2.0 /graphify skill for Claude Code, Cursor, Codex and Gemini CLI that maps a whole project, including code, docs, SQL schemas, configs and PDFs, into a knowledge graph the agent can query instead of grepping through files. Code is parsed locally with tree-sitter, so the map is deterministic, needs no LLM or vector store, and nothing leaves your machine, and every edge in the graph says why it exists. A separate paid Graphify Cloud adds always-on team memory with a 14-day trial.; Archify: Archify is an Apache-2.0 agent skill for Cursor, Claude Code, Codex CLI and OpenCode that turns an idea, plan or whole codebase into an interactive HTML diagram you can explore, follow source links through and share, from repository maps to learning paths and travel plans.
Are there free or open-source options?
Yes: 32 open source, 2 source-available, 3 free or freemium, 2 paid, usage-based or enterprise, 2 pricing on the maker's site. Pricing is as each maker states it.
More from writing code
Tools are grouped by what each says it does. Missing one? Submit it free.