Freesolo CLI

Interactive command-line client for Freesolo training jobs.

The CLI:

• prompts for setup input
• verifies a Freesolo API key
• enqueues /api/training/jobs
• polls /api/training/jobs/{job_id} until the worker finishes

Install

pip install freesolo-agent

This installs the freesolo command.

Setup

Run:

freesolo setup

The CLI prompts for:

• Freesolo API key
• operation: draft, optimize, training, or poll
• source repository URL for draft
• the latest stored organization job to continue for optimize and training
• optional branch and worker environment variables for draft

For private repositories, install the Freesolo GitHub App when prompted:

https://github.com/apps/freesolo-agent/installations/new

The CLI prints progress while the backend job is queued and running. When the job finishes, it prints a short summary with the job id, repo, commit, and changed files when available. Full job details and richer summaries live in the Freesolo platform at https://freesolo.co.

Optimizer jobs use the synchronous OpenAI oracle path. Training jobs use the batch-backed SFT labeling path.

Operations

draft creates the initial Freesolo training contract from a source repo. It returns a targetRepoUrl for the generated training repo.

optimize generates repo-level training files and keeps running bounded strategy-discovery experiments for the latest job in a draft chain. One optimize job is instructed to run at most 5 heavy experiments concurrently, with each experiment capped at 1 hour.

training tunes the optimizer-selected SFT and RL script/config combination for the latest job in a draft chain after that chain has a successful or user-stopped optimize job.

poll lists recent organization jobs and watches the selected job until it finishes. It does not enqueue new work.