Cadre

Search Cadre docs

Find guides, workflow references, and architecture pages.

Search Cadre docs

Find guides, workflow references, and architecture pages.

Install

Cadre

Measure twice, code once.

Cadre is a context-driven development harness for AI coding agents, combining spec-first tracks, Beads-backed task memory, review gates, team boards, parallel worker orchestration, and mono/polyrepo delivery.

Get startedRead the architecture
Claude Code
OpenAI Codex
Beads memory

Packet-owned workflow

Cadre MCP coordinates every state change.

01

Setup

02

Track

03

Implement

04

Review

05

Ship / Land

06

Archive

MCP
Beads
LSP

A workflow layer agents can actually operate.

The docs are organized around how Cadre works in practice: setup, planning, implementation, review, delivery, teams, and internals.

Packet-owned workflows
Agents call deterministic Cadre packets instead of editing plans, metadata, Beads, or review state by hand.
Durable task memory
Beads keeps the task graph, dependencies, notes, blockers, and handoffs available across sessions.
Team-scale delivery
Ownership, advisory leases, review queues, provider evidence, and mono/polyrepo publication stay coordinated.

Documentation map

Start with the workflow guide, then go deeper into architecture, team-scale operation, parallel execution, and support.

Getting Started
Open guide
Architecture
Open guide
Team + Polyrepo
Open guide
Troubleshooting
Open guide

All guides

Markdown-backed pages rendered by the Next.js docs shell.

8 pages
Overview
Cadre
Context-driven development harness for AI coding agents.
Read
Start
Getting Started
Install Beads, install the Cadre plugin, and initialize a target project.
Read
Core Concepts
How Cadre Works
Packet-owned workflows, MCP runtime, Beads memory, review gates, provider evidence, and code intelligence.
Read
Core Concepts
Workflows
Detailed guide to the Cadre workflow lifecycle and every cadre-* command.
Read
Internals
Architecture
Harness package layout, generated plugin bundles, source files, and development flow.
Read
Scale
Team And Polyrepo
Shared sync, ownership, leases, team boards, polyrepo control repos, and merge trains.
Read
Scale
Parallel Execution
Phase annotations, worker waves, file claims, merge-back, and failure recovery.
Read
Support
Troubleshooting
Common install, MCP, Beads, provider, LSP, and generated-bundle failures.
Read