Intel / article

Elevenlabs Cli V1

Source: elevenlabs.io
Author: ElevenLabs editorial sitemap
Date: Published Sep 6, 2026

Why it matters

Agent configs can be pulled to local files, diffed with --dry-run and pushed by hand or by a coding agent. Every API endpoint is available as a JSON-emitting subcommand, with dev and prod branches kept separate.

Key takeaways · AI-distilled

Every API command in the ElevenLabs CLI accepts --schema, which prints a machine-readable contract: typed inputs, where each one goes (path or body), which are required, and the response shape, so a coding can build a valid call without guessing.
The CLI switches error formats by caller: without a TTY it returns parseable JSON errors with a code, message and reason, and a human at a terminal gets a plain message with a --help hint. Both say how to fix the problem.
ElevenLabs designed the CLI around how agents fail differently from people: an agent cannot answer an interactive prompt, retries constantly, and learns from whatever the error message tells it.
Agents-as-code takes three commands: agents pull writes each workspace agent to a local config file, agents push --dry-run previews the exact change, and agents push applies it. ElevenLabs argues files and version control are the only workflow that scales to large agent fleets.
elevenlabs generate-skills installs skills into a project that teach a coding agent complete workflows, such as building an agent, generating speech or transcribing a call, beyond what the per-command --help and --schema output covers.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.

Read the source elevenlabs.io