> ## Documentation Index
> Fetch the complete documentation index at: https://docs.reactor.inc/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> To build and serve your own model, start at /deploy/development/quickstart and /deploy/development/overview. Deploying is the default path: reactor init scaffolds a workspace, reactor auth login authenticates, and reactor model deploy registers the model, publishes the release with the weights/ folder, and activates it on Reactor's GPUs, in one command from that workspace. Docker must be running, because the publish step builds the image locally. Bump model.version in reactor.yaml before redeploying a change, because a release that already has an image is reactivated as it is. Deployment access is granted per account, so contact team@reactor.inc if a deploy is refused. Every key in reactor.yaml is documented at /deploy/platform/reactor-yaml. Model code imports reactor_runtime; Python client code imports reactor_sdk. The runtime overview explains the model interface. Running the model on your own machine with reactor run is optional and needs a GPU you attach with --gpus; /deploy/development/local-testing covers that loop and pairs a complete brightness model with a Python client test in a separate brightness-test workspace.
> Reactor hosts multiple models, each with its own connect slug (modelName) and command/event schema. The video model catalog — slug, typed SDK package, and links to its schema — is at /model-api-reference/overview. Robotics policy documentation starts at /robotics/overview; X-WAM observations, actions, and client integration are under /robotics/xwam/; Cosmos3 Nano Policy DROID is under /robotics/cosmos/nano-policy-droid/. Some models expose one slug per experience (e.g. HappyOyster); always take the slug from the model's own pages, never guess it.
> Fastest path to a working app: `npx create-reactor-app my-app --model=<slug>` scaffolds a complete app with secure auth wired up. Typed TypeScript SDKs are published as @reactor-models/<model>; Python uses the base reactor-sdk package.
> Auth: exchange an API key (rk_...) for a JWT via POST https://api.reactor.inc/tokens from your server. Never put the API key in client-side code.
> Append .md to any docs URL for clean Markdown. Search these docs via the MCP server at https://docs.reactor.inc/mcp.

# Troubleshooting

> Diagnose missing replies, observation mismatches, and action interpretation errors.

| Symptom | Check |
| - | - |
| Repository returns 404 | Use the public `reactor-team/reactor-cookbook` URL and the main-branch installation instructions in the quickstart. No GitHub organization access is required. |
| Session creation returns `429` with `no available capacity` | There is no available model capacity for this request. Retry later or contact Reactor; changing observation fields will not resolve it. |
| Session never reaches `READY` | Inspect SDK status and error logs. Confirm the API key, model access, and endpoint; the reference client allows 300 seconds for readiness. |
| Session is ready but no actions arrive | Set a nonempty task and keep all three correctly named tracks publishing after the request. Verify `proprio` contains 16 finite numbers and `state_json` is a JSON string. |
| Re-sending a request gives no reply | An unchanged `state_json` is deduplicated. Preserve the request ID and seed fields, and increment `retry`. |
| Replies belong to old observations | Keep only one request outstanding. Check the `step` echo, discard stale replies, and allow updated frames to clear the encoder before sending state. |
| Actions look plausible but move incorrectly | Check camera assignments, RGB order, `wxyz` quaternion order, coordinate transforms, gripper scale, and sequential delta integration. A valid response does not validate the robot configuration. |
| Reported latency is lower than application latency | The client's timer excludes frame settling, capture, and action execution. Instrument those separately. |
| Simulator cannot import `X-WAM` | Copy `evaluation/X-WAM` into `third_party/RoboTwin/policy/` and launch from the X-WAM repository root. |

Malformed state can be dropped without an action reply. Validate inputs locally instead of treating
every timeout as a transient network error. See the
[wire contract](/robotics/xwam/robotwin/reference) for field limits and ordering.

When reporting a problem, include the model identifier, client revision, SDK version, request ID,
status transitions, input shapes, and timing. Keep credentials out of logs and reports.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.