Integration

pyghidra-mcp

PyGhidra-based MCP server that analyzes every binary in a Ghidra project and exposes decompilation, cross-references, call graphs, symbol and string search, indexed semantic code search, and renaming, typing and comment edits. A separate HTTP CLI client and an optional mode driving a Ghidra GUI it launches are included.

Requirements and limitations

Reviewed release: v0.2.7, published 2026-09-25, with package metadata requiring Python >=3.10. The README describes the project as beta and under active development; main has later, unreleased MCP Python SDK 2 changes.
Outside the published container image, install Ghidra separately and point GHIDRA_INSTALL_DIR at it. Semantic search uses a local ChromaDB index built in the background, so it can lag behind decompilation on large projects; ChromaDB and GZF artifacts are stored beside the Ghidra project.
The CLI client and GUI mode require the streamable HTTP transport; GUI mode launches Ghidra in-process and cannot attach to an already-running Ghidra. Import, delete, rename, type and comment tools change the project.
Apache-2.0. No server, client or analysis was run.

About this entry

Type: Integration
Category: Independent MCP integrations
Relationship: Independent agent integration
Targets: Binary

Sources reviewed 2026-10-17

The description was reviewed against primary sources. This does not establish runtime testing, security auditing or verified compatibility.

Requirements and limitations

About this entry

Primary sources