OpenParable
Skip to content
← Back to skills

Muesli Agent

Apps & Automation
muesli-agent

Muesli Agent is an Agent Skill for apps & automation published in the Muesli-HQ/muesli repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Use when working with local Muesli meetings, notes, dictations, audio-file transcription, or raw transcripts through the bundled `muesli-cli` CLI. Prefer this skill when a coding agent needs to transcribe local audio, inspect transcripts, summarize meetings with its own model, or write notes back into Muesli without requiring the user's API keys.

View on GitHub ↗
QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

  1. Open your project in Codex.
  2. Copy and paste the prompt below.
  3. Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/Muesli-HQ/muesli/blob/HEAD/skills/muesli-agent/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/muesli-agent, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Muesli Agent

Use the local muesli-cli CLI as the source of truth for meeting and dictation data.

CLI discovery

Resolve the binary in this order:

  1. command -v muesli-cli
  2. command -v muesli only when the resolved path is a Homebrew cask alias to Muesli.app/Contents/MacOS/muesli-cli; verify with muesli info
  3. /Applications/Muesli.app/Contents/MacOS/muesli-cli
  4. A local SwiftPM build path inside this repo

If discovery is uncertain, run the candidate binary with info first and reject unrelated muesli executables.

Core workflow

  1. Inspect capabilities with muesli-cli spec if you do not know the exact subcommand shape.
  2. To transcribe a local audio file, run muesli-cli transcribe <file> and read plain transcript text from stdout.
  3. Use muesli-cli transcribe <file> --format json when you need duration, word count, model, warnings, summary, or saved meeting ID.
  4. Add --save-meeting to persist an imported meeting with source = audio_import.
  5. List candidate meetings with muesli-cli meetings list --limit 10.
  6. Fetch a full record with muesli-cli meetings get <id>.
  7. Use the coding agent's own model to analyze rawTranscript and formattedNotes.
  8. If you want to persist improved notes, write markdown back with:
    • cat notes.md | muesli-cli meetings update-notes <id> --stdin
    • or muesli-cli meetings update-notes <id> --file notes.md

Rules

  • Treat CLI stdout as the API. Data commands are JSON by default; transcribe is plain text by default, Markdown with --format markdown, and JSON with --format json.
  • Treat stderr as informational only.
  • Do not mutate rawTranscript; only update formattedNotes.
  • Prefer the meeting transcript when notesState is missing or raw_transcript_fallback.
  • Use --db-path or --support-dir only when the default Muesli data location is wrong.

When to read references

Read references/cli-contract.md if you need the exact command tree, field definitions, or failure behavior.