Hexmux

Hexmux is a small local control plane for agent-driven IDA Pro. Agents send task-specific IDAPython scripts through a per-user Unix-socket supervisor; IDA connects outward to that supervisor and never opens a network port.

Development setup

cd hexmux
python3 -m pip install -e .
mkdir -p ~/.idapro/plugins
cp plugin/hexmux_plugin.py ~/.idapro/plugins/

Restart IDA. The plugin asks the installed hexmux CLI to start the supervisor, then connects in the background. If IDA's GUI environment cannot find the executable, set HEXMUX_EXECUTABLE to its absolute path before launching IDA.

Related MCP server: IDA Pro MCP

Use

The plugin or the first CLI command starts the detached supervisor on demand:

hexmux status
hexmux ps

Generate a script:

import idautils
import ida_name

result = [
    {"ea": hex(ea), "name": ida_name.get_name(ea)}
    for ea in list(idautils.Functions())[:20]
]

Then run it using the ID or unique prefix printed by hexmux ps:

hexmux run 7c91a42b functions.py
hexmux run 7c91a42b --json < functions.py

result should normally be JSON-compatible. Printed output and exceptions are captured separately. Use --json when another program or agent will consume the response.

MCP

Configure an MCP client to launch:

hexmux mcp

The stdio server exposes seven tools: instances, python, decompile, disass, xrefs, search, and annotate. Every IDA-facing tool requires an explicit instance selector. python is the general interface and returns the value of a trailing expression; the remaining tools are compact conveniences implemented as generated IDAPython scripts over the same execution protocol.

See DESIGN.md for protocol and MCP adapter decisions.

Fetch IDAPython reference stubs

Fetch a pinned, source-form documentation snapshot without retaining the SDK checkout:

python3 scripts/fetch_idapython_reference.py \
  --ref v9.3.0-release \
  --output references/9.3

This copies the curated apidoc/*.py files as .pyi documentation stubs, plus the qualified symbol inventory and examples index. These curated stubs are useful but not a complete substitute for runtime-generated IDAPython documentation.