Mache parses codebases to discover their structure and exposes it as a navigable graph via MCP tools or a filesystem mount. It allows agents to explore call chains, jump to definitions, and read context without grepping flat files. Built for AI agents, it provides stable node paths and identity-preserving write-back.

Is mache open source?

Yes, licensed under Apache-2.0.

mache: Parses codebases to discover their structure and exposes it as a… (★ 41, Apache-2.0)

Features

01Tree-sitter parsing (28 languages)

02MCP server (17 tools, stdio + HTTP)

03Cross-references (callers/callees)

04NFS mount + write-back

05Schema inference (FCA)

Why choose it

+Tree-sitter parsing (28 languages)

+Agent-based code exploration and navigation

+Ships with a public repository and a Apache-2.0 license, which makes adoption and review easier.

+The latest recorded update is 2026-07-17, which suggests the project is still actively maintained.

Trade-offs

!Compatibility coverage looks narrow, so it is worth validating your target runtime before committing to it.

!There are at least 8 related tools in the same category, so the best choice is easier to make after side-by-side comparison.

Compatibility

Linux

Verified via docs

macOS

Verified via docs

Quick start

$ git clone https://github.com/agentic-research/mache.git

$ cd mache

$ task build

$ task install

Use cases

↳Agent-based code exploration and navigation

↳Automated code smell detection

↳Structured data schema inference and traversal

How it compares

≈mache sits in the Data Processing category, so it makes more sense to evaluate it alongside tools like mirascope instead of in isolation.

≈If your main need is closer to "Agent-based code exploration and navigation", that use case is a better lens for comparison than broad feature checklists alone.

≈mache uses a Apache-2.0 license, and community traction are both easier to judge in category context.

Alternatives

mirascope★ 1.5k

LLM abstractions that aren't obstructions

vs →

FinanceToolkit★ 5.1k

Transparent and Efficient Financial Analysis

vs →

pmb★ 313

Local-first persistent memory for AI coding agents (Claude Code, Cursor, Codex) via MCP. 94.5% LoCoMo recall@10, 70ms p50, multilingual, zero API keys.

Comments

?
usr_seed_0167Apr 26, 2026
The IDE-style navigation metaphor is well-chosen, the experience feels familiar
?
usr_seed_0201Apr 23, 2026
IDE-style navigation for JSON and YAML data is a practical improvement over text search
?
usr_seed_0237Apr 23, 2026
Jump-to-definition semantics for structured data reduces exploration time significantly
?
usr_seed_0037Mar 31, 2026
Used for navigating large configuration files and API responses, very intuitive

mache