LLM Reference Docs
Two machine-readable files are available at well-known URLs, following the llmstxt.org standard:llms.txt
Link index — one line per doc page. Use when you want an agent to discover and fetch individual pages.
llms-full.txt
Full reference — the entire API in one document, with canonical flows, gotchas, and code examples. Fetch once and the agent knows the full API.
Skills
The Bota Skills repository contains customer integration recipes for AI assistants. The customer repository is public and MIT-licensed; reading it requires no organization membership. A published installable skills bundle and assistant discovery remain separate. The ten recipes are source-reviewed against pinned API and package contracts. Catalog integrity and synthetic webhook checks do not certify live services, physical provisioning or package availability. Use the API reference, CLI guide, and MCP guide to check supported operations and releases. Skills provide task context; they do not grant permissions. Keep API keys and tokens out of model prompts.Device Registration
Registration, customer identity mapping, binding and reassignment
Recording Upload
Upload contracts, verification and recovery boundaries
Webhook Handling
Signature headers, signed payloads and delivery handling
Auto-Processing
Configuration routes, schemas and processing admission
Streaming Upload
Streaming fields, chunk routes and supported App SDK workflows
API SDK
Experimental TypeScript and async Python clients, supported operations and release boundaries
CLI
Explicit origins, API-key profiles and authorized terminal workflows
Local MCP
Noninteractive stdio setup and local compatibility contracts
Hosted MCP
Direct HTTP or stdio bridging, authorized discovery and uncertain outcomes
Shared Authentication
Project credentials, profile custody, end-user enrollment and Platform enrollment boundaries
MCP Server
The Bota MCP server connects your AI assistant directly to your Bota project — list devices, inspect recordings, trigger test webhooks, and more without leaving your IDE. You can also issue scoped tokens so end users can query their own recordings with an AI assistant.MCP Server
Developer tier (full project access) and end user tier (scoped to their recordings)

