Reaper-MCP lets AI assistants edit audio tracks

What this is about

Reaper-MCP is an open-source bridge between MCP-compatible AI assistants and the REAPER digital audio workstation. Instead of merely discussing audio editing, a connected assistant can send concrete commands to an open project: create tracks, insert media, read markers, configure effects, edit MIDI, or retrieve measurements.

Version 0.7.1 was released on September 15, 2026. It fixes, among other issues, unbounded result sets in large projects and a path-safety flaw in recursive folder scans. That makes the tool relevant to people who want repeatable audio-production automation without replacing the DAW itself.

What Reaper-MCP actually does

According to its documentation, the project exposes 172 tools. A Python MCP server receives structured requests and communicates locally with a ReaScript module inside REAPER. Available areas include project and track management, media items, effects, automation envelopes, MIDI, markers, analysis, rendering, batch processing, and composition helpers.

The separation between language and action is important: the language model chooses which tool to call and with which parameters. REAPER performs the actual audio operation. Project files, plug-ins, and playback therefore remain in an established production environment. Profiles can limit the exposed tool set for jobs such as production or post-production.

The current version does not add a flashy new interface; it strengthens operations. Large lists are capped earlier, folder scans no longer follow symbolic links outside the permitted root, and known credential-related directories are excluded. Those details matter because an agent could otherwise see excessive project data or files outside the intended workspace.

Why it matters

Audio editing often consists of many small, exact steps: mark every breath pause, normalize twelve clips, inspect noise, apply an effect to selected tracks, and export several variants. A language interface is most useful where those steps can be named clearly and repeated under control.

For producers, podcast teams, and game-audio departments, Reaper-MCP can act as an automation layer. The assistant can read project state and combine actions while a person still listens, decides, and approves. The Apache 2.0 license permits custom modifications. Users still need a working REAPER installation, suitable plug-ins, and an MCP client; the language model is not included in the repository.

In plain language

Reaper-MCP is like a studio assistant with a precise catalog of controls. Instead of only saying “clean up the podcast,” the assistant can operate the mixer, scissors, and markers through defined handles. The audio engineer remains at the console, listens to the result, and can correct every step.

A practical example

A podcast team receives a 55-minute recording with two voice tracks every week. After import, the assistant should name the tracks, place 18 known markers, find loud peaks, apply a prepared effect chain, and create three render regions. Reaper-MCP translates that job into individual tool calls and then reads the project values again.

The editors listen to the transitions and approve the final render only afterward. If the workflow saves just 20 minutes per episode, that is 80 minutes of recurring manual work across four monthly episodes. A sensible trial starts with a copy of a small project: read-only access and markers first, reversible edits second, and controlled rendering last.

Scope and limits

• The language model can misunderstand a task or choose unsuitable parameters. Backups, restricted permissions, and human review are essential before destructive changes.
• Sound quality cannot be inferred from project values alone. Mixing, timing, and aesthetics still require trained listening.
• Plug-ins, operating systems, and REAPER configurations differ. A workflow that works on one computer may not behave identically on another.
• The connection expands the attack surface. Version 0.7.1 fixes concrete path issues, but users should connect only trusted clients and keep the permitted file scope narrow.

SEO & GEO keywords

Reaper-MCP, REAPER, Model Context Protocol, MCP, Audio Automation, Music Production, Podcast Editing, ReaScript, MIDI Editing, Audio Postproduction, AI audio tool