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

Skip to main content

ctrlrun home page

How it works
GitHub
Protect your first action
Protect your first action

Start

Overview
Why
Not only agents
Agents you can't modify
The boundary

Get started

Install
Protect your first action
Three ways in
Choosing between them

Production

Run it in production
SQLite or Postgres
Running on Postgres
How reservation works
Migrations and schema versions
Recovery after a crash
Receipt integrity in practice
Anchoring the receipt chain
Retention: pruning the receipt chain
The soak, and what it does not establish
Operations

Concepts

Action and hash
Decisions
Approval binding
Effect keys
Outcomes and AMBIGUOUS
Receipts and evidence
Authority and delegation
Authority in depth
Observe mode
Fail closed

Guides

Protect a function
Put the gateway in front of MCP
Approve in Slack
Resolve an AMBIGUOUS effect
Reconcile automatically
Roll out observe, then enforce
Run on Postgres
Verify in CI
Export to OpenTelemetry
Adapters
Use the LangChain middleware
Use the LangGraph adapter
Use the OpenAI Agents SDK adapter

MCP

Overview
The gateway in five minutes
Approve from your assistant
Use the docs from your editor

Cookbook

Cookbook

Verify

ctrlrun verify
What verify guarantees
Get the verified badge

Compare

vs framework HITL
vs guardrail libraries
vs oversight toolkits
vs durable workflows
vs idempotency keys
Double-execution study

FAQ

FAQ

THE EXECUTION BOUNDARY

The model decides what.
ctrlrun decides whether.

Pick your domain. The picture is the whole decision.

Give agents autonomy.
Keep control of their actions.

ctrlrun ↗Quickstart ↗GitHub ↗