Installation

Install batchwork-ai with uv or pip, set provider credentials, and verify your environment before submitting the first batch job.

With uv:

uv add batchwork-ai

With pip:

pip install batchwork-ai

Batchwork requires Python 3.11 or newer. The base install contains HTTPX and Pydantic; provider SDKs are not runtime dependencies.

Install the CLI

The same package ships the batchwork command, so you can submit and manage batches from the terminal without writing Python. Install it as an isolated tool:

uv tool install batchwork-ai
batchwork --version
batchwork --help

To try it once without installing anything permanently:

uvx --from batchwork-ai batchwork --help

Help, version, and shell completion work without configuration, a registry, or credentials. See the CLI guide for human and agent quick starts.

Add the optional Upstash Redis store only when needed:

uv add "batchwork-ai[redis]"
# or
pip install "batchwork-ai[redis]"

Continue with Configuration for model syntax, provider credentials, endpoint overrides, supported modalities, and batch limits. Credentials may be configured once on Batchwork, supplied per call, or read from environment variables.

Submitting a batch performs network requests and creates a real provider job. Provider usage and storage charges may apply.

Install the Agent Skill

If you use a coding agent such as Claude Code or Cursor, the batchwork-ai Agent Skill teaches it to drive the CLI safely: when to ask before spending money, how to keep job identity across turns, and how to recover interrupted work.

npx skills add ajanraj/batchwork-ai@batchwork-ai