Voice Forge

A minimal CLI for managing ElevenLabs Conversational AI configs with Git.

Quick Start

1. Install the CLI

   • npm install -g voice-forge

2. Create a project

   • mkdir my-assistant && cd my-assistant
   • vf init

3. Configure your environment

   • cp .env.example .env
   • Add your API key(s) to .env (do not commit .env)

4. Workflows

   • Check status: vf status
   • Pull remote configs: vf pull
   • Dry‑run a deploy: vf deploy --dry-run
   • Deploy: vf deploy

Project Layout (default)

project/
├─ agents/
│  ├─ defaults.json
│  └─ agent-one/config.json
├─ tools/
│  └─ tool-one.json
├─ knowledge/
├─ dictionaries/
├─ .forge/
│  ├─ config.yml
│  └─ state.json (git-ignored)
├─ .env (git-ignored)
└─ .env.example

Basics

• Use human‑readable names in configs; the CLI resolves names ↔ IDs for you.
• Shared settings go in agents/defaults.json.
• Secrets live in .env and are synced to ElevenLabs (not stored in configs).
• Production actions usually require confirmation flags.

Common Commands

• vf init — scaffold a project
• vf status — show local changes and shared config impact
• vf pull — fetch remote resources to local files
• vf deploy — apply local changes to ElevenLabs
• vf secrets — manage secret mappings and values

Notes

• Keep .env and .forge/state.json out of version control.
• Examples in this repo are generic and not tied to any specific business.

For detailed usage, run vf --help or vf <command> --help.

Security

• Do not commit secrets; keep .env out of version control.
• Prefer --dry-run and confirmation flags for sensitive or production actions.
• Keep secrets out of config files; use secret mapping and sync commands instead.