agent-rollouts 0.1.4

A CLI to get training data from your own coding agents. Track rollouts and codebase snapshots at every turn. The data can be used for SFT, RL, and Continual Learning.

Rollouts

A CLI for Continual Learning with your own coding agent sessions. Track rollouts and codebase snapshots at every turn. The data can be used for SFT, RL, and Continual Learning. Currently supports OpenCode and uses Prime Intellect for hosted RL training.

Quick Start

# Install rollouts
uv tool install agent-rollouts

# Install the OpenCode Rollouts plugin.
# If GitHub repo defaults are not configured yet, setup will prompt for them.
rollouts setup

# Start a learn session
rollouts learn start my-session --config /path/to/rl.toml

# Continue from the latest run, reusing its config and the latest available checkpoint
rollouts learn continue my-session

# Continue from a specific run with a different config file
rollouts learn continue my-session --from-run 2 --config /path/to/another-rl.toml

# Restart the latest run in the same session if it failed or was stopped
rollouts learn restart my-session

# Or override the config path when restarting
rollouts learn restart my-session --config /path/to/rl.toml

# Inspect learn sessions
rollouts learn list
rollouts learn status my-session

Make sure prime, opencode, git, gh, and hf are installed first.

If you plan to use rollouts learn, make sure your Prime environment is also configured.

Installation

Using uv

Install from PyPI:

uv tool install agent-rollouts

Upgrade later:

uv tool upgrade agent-rollouts

Usage

Learn Sessions

• rollouts setup installs the OpenCode Rollouts plugin and prompts for GitHub repo defaults if they are not configured yet.
• rollouts remote defaults set is required before rollouts learn start because learn sessions push tracked snapshots and may auto-create GitHub repos.
• rollouts learn start <session> --config <path> creates the learn session, stores the config, syncs the dataset, creates run #1, and starts the initial Prime run.
• rollouts learn list shows your learn sessions.
• rollouts learn status <session> shows the latest run, current Prime state, config path, and run lineage.
• rollouts learn restart <session> restarts the latest failed or manually stopped run in the same session after confirmation.
• rollouts learn restart <session> --config <path> overrides the stored config path for the restarted run.
• rollouts learn continue <session> starts a new child run from a previous run, reusing the stored config and the latest available checkpoint by default.
• rollouts learn continue <session> --from-run <n> continues from a specific run instead of the latest run.
• rollouts learn continue <session> --checkpoint none starts the new run without setting checkpoint_id.
• rollouts learn continue <session> --checkpoint <id> warm-starts from a specific Prime checkpoint id.

If an older run was created before config-path tracking existed, rollouts learn restart may require --config the first time you retry it.

Local Repo Setup

For local development:

uv sync --extra dev

To install the current checkout as a tool:

uv tool install --editable .

You usually do not need to pass --python here as long as uv selects a compatible interpreter for the project.

Development

Quality checks:

uv run ruff check .
uv run ty check src

Pre-commit hooks are configured for:

• ruff --fix
• ruff format
• ty check src/

Install the Git hook locally with:

uv run pre-commit install

Run hooks manually with:

uv run pre-commit run --all-files