Guides & Cookbooks

Base URL, API key, model name. Works with Cursor, opencode, LangChain, and any OpenAI-compatible tool.

The difference between a model and an intelligence layer. Outcomes, not features.

Install the CLI. Chat, manage memory, test policies, and see what happened from your terminal.

The /v1/chat/completions endpoint. Streaming, tool calling, reasoning content, what's different from OpenAI.

The recommended endpoint for new projects. Governance, memory, tool execution, and conversation state in one call.

Create rules, check what it does, understand the approval flow. Every action checked.

Hybrid execution. Your tools pass through. Cadreen's tools execute with governance.

Create, configure, deploy, and manage agents. Versioning, health monitoring, and lifecycle.

Connect workspaces. Cross-workspace agent collaboration with explicit consent and governance.

Agent-to-agent protocol. Connect external agents via A2A with governance-gated task execution.

Register Cadreen as an MCP server. Connect governance, memory, and learning to any MCP-compatible client.

Blueprints are reusable templates. Create one from a trace, run it with new parameters, automate it with schedules.

Attach a schedule to any blueprint — daily, weekly, or cron. Pausing, resuming, and webhook events included.

Write a policy in natural language. Cadreen enforces it across every surface — opencode, SDK, CLI, dashboard.

Change two fields in your existing OpenAI-compatible code. Same messages, same tools — with intelligence added.

Text generation, multi-turn conversations, tool calling, streaming, and structured output with the Responses API.

The system blocks the tool call, returns a conversational prompt. Human says yes. No custom approval UI needed.

When tools fail, Cadreen figures out what went wrong, fixes arguments, retries. Remembers the fix for next time.

Upload an OpenAPI spec. Cadreen auto-generates managed tools — ready to use, with error handling and health checks.

Every response carries a trace ID. Fetch the full decision breakdown. Tamper-evident for compliance.

Cadreen spots repeated patterns and proposes turning them into reusable workflows. You approve. The system gets smarter.

The system spots patterns in your usage — recurring actions, time-based routines, automation opportunities — and proposes them as tasks. You decide what runs.

Go SDK in GitHub Actions. Cadreen assesses risk, checks rules, decides: deploy, asks for permission, or blocks.

One workspace, five surfaces, zero sync code. A fact stored in one surface appears everywhere immediately.

Pre-built compliance policy templates. One call. Your workspace enforces the standard across every tool call.

Create an agent, connect services, set policies, and deploy — all from the dashboard or CLI.

Set up federation between two workspaces. Agents share knowledge across organizational boundaries.

Register an external agent via A2A protocol. Governance-gated task execution across platforms.

Create agents from scratch or templates. Configure capabilities, set policies, deploy, and manage versions.

Cadreen assesses risk in your deployment pipeline, checks rules, and decides: deploy, asks for permission, or blocks.

One workspace, five surfaces, zero sync code. A fact stored in one place appears everywhere immediately.