Guides
Task-oriented guides built from runnable samples.
Every guide in this section is built from a runnable sample in the repository's
samples/ directory.
Each one walks through a working flow YAML, explains the pieces as it builds
them up, and gives you the exact commands to run and test it. Most samples need
zero setup beyond the octo binary.
HTTP & APIs
Build a REST API
Serve HTTP endpoints with routing, path params, and response shaping.
Call External APIs
Use the HTTP client with auth, retries, and response caching.
Database CRUD
Query SQLite or Postgres with parameterized SQL.
MongoDB Documents
Store and query JSON documents, and total them with an aggregation.
Data
Transforming Data
Reshape payloads with multi-transform, enrich scopes, and CEL.
Splitting and Aggregating
Process a collection element by element, then re-join the results.
Validation and Auth
Validate input with CEL rules and protect endpoints with JWT.
Verifying Webhooks
Authenticate a signed webhook from any provider with HMAC in CEL.
Encrypting Data
Keep a field secret through a queue, a column or a log line, and read it back.
Debugging a Flow
Stop at any block and inspect the message, or learn the block was never reached.
Testing Flows
Write a flow's tests beside it, mock what calls the world, and run them in CI.
Composition & messaging
Composing Flows
Call flows from flows: flow-ref, sourceless flows, and HTTP composition.
Events and Queues
Broadcast with topics or load-balance work with queues.
Scheduled Jobs
Run flows on a cron schedule and write structured logs to files.
State
Object Storage and Caching
Persist state in the object store and memoize with cache scopes.
Reading and Writing Files
Read and write files under a directory you name, with a path the caller picks.
Web
Integrations
Build a Slack Bot
Receive Slack events, verify signatures, and reply to mentions.
Work with Notion
Handle Notion webhooks, query data sources, and render pages as Markdown.
Async Research with Parallel
Start a long-running research job and take its answer back as a verified webhook.
Running Commands
Run a local program, stream its output, and hand a model a safe toolbox.
AI
An AI Web App End to End
Cache LLM output and render it as HTML with an error fallback page.
Build a RAG Pipeline
Embed with Gemini, search in Pinecone, and let an agent answer from the matches.
Human in the Loop
Make an agent ask a person before it runs a tool, and handle nobody answering.
Agentic Self-Healing
Turn a platform alert into an agent that triages it and fixes it with permission.
The AI section
The full AI documentation: agents, memory, skills, routing, and MCP.