Agent discovery order

Agents should start with machine-readable site discovery, then use the repo-local ota.yaml when operating inside a checkout.

Discovery sequencetext

https://ota.run/llms.txthttps://ota.run/openapi.jsonhttps://ota.run/.well-known/agent-skills/index.jsonhttps://github.com/ota-run/skillshttps://ota.run/docs/index.json

Repository operation order

Inside a repository, agents should prefer ota commands over inferred package-manager commands when the repo exposes an ota contract.

Safe repo loopbash

ota doctorota validateota tasks --safe --useota up --dry-runota run <task> --agent

Boundaries agents must preserve

treat ota.yaml as the source of truth for repo readiness
respect declared safe tasks, writable paths, setup requirements, and verification commands
use JSON output when automation needs stable machine-readable status
prefer execution receipts when a task result needs auditability
do not invent contract fields, hidden setup steps, or parallel workflows outside the canonical docs

Integration surfaces

OpenAPI

Use the OpenAPI spec to discover public ota.run resources and standards endpoints.

When to use it

Use this for API catalogs and agent tooling that understands OpenAPI.

Example

Official Ota skill

Use ota-run/skills for Ota-specific agent instructions around contract authoring, contract review, readiness modeling, and Studio boundaries.

When to use it

Use this for coding agents that support skills or need a canonical operating guide.

Link

Agent skills

Use the skills index to find canonical commands, docs, and registry links for repo readiness, task execution, and CI.

When to use it

Use this for coding agents that need a compact action map.

Example