Source for docs.cowboy.inc and the example actors that go with it. Cowboy is a Layer-1 for autonomous agents: Python actors run in a metered VM on validators, keep persistent state, schedule their own work, and call off-chain services (LLMs, HTTP, MCP, containers) through runners.
| Directory | What it is |
|---|---|
docs/ |
The documentation site (Mintlify). Start at docs/getting-started/quickstart.mdx. |
examples/core/ |
The numbered builder curriculum: one actor per lesson, in order. |
examples/gallery/ |
Product-shaped demos that combine the core patterns. |
examples/README.md |
Which lesson teaches what, and how to run the sweep. |
Install the CLI, then run any lesson against the hosted devnet:
brew install cowboyinc/tap/cowboy
cowboy versionEach lesson directory has its own README with the exact commands. The runtime (node, runner, storage, secrets) is not in this repository; the CLI and the hosted network are the interface.
This tree is promoted from Cowboy's private working repository on a regular
cadence, as a snapshot without history. PROMOTED_FROM records the source
commit. Corrections are welcome as issues or pull requests here; accepted
changes are applied at the source and arrive with the next promotion.