What is conversation compaction in Claude Code?

Compaction occurs when an AI's context window reaches its limit, requiring the history to be summarized or reset. Beads stores task state in a git-backed graph, allowing Claude to recover full context immediately after compaction.

Can I use Beads for team collaboration?

Yes. Because Beads is Git-backed, the task graph can be synchronized across a team using 'bd sync', allowing both humans and AI agents to work on the same issue set.

How does Beads differ from the TodoWrite tool?

TodoWrite is designed for simple, linear, single-session checklists. Beads is built for complex, multi-session work involving dependencies, blockers, and team collaboration via Git.

Does this skill require external software?

Yes, it requires the 'bd' CLI (v0.47.0+) to be installed in your PATH and a Git repository to be initialized for the project.

Beads Persistent Task Manager

Name: Beads Persistent Task Manager
Author: mbarnes-code

bymbarnes-code

生产力与工作流

Manages persistent, graph-based tasks and dependencies that survive AI conversation compaction and span multiple work sessions.

关于

Beads is a git-backed issue tracker designed specifically for long-running AI agent workflows where context needs to persist across multiple sessions or after conversation compaction. Unlike ephemeral checklists, Beads maintains a graph-based structure for complex task dependencies, allowing Claude to track blockers, resume work days later, and sync progress with human teams via Git. It is the essential tool for managing non-linear engineering tasks that exceed the memory limits of a single chat session.

主要功能

Git-backed persistent memory for multi-session context recovery
AI-optimized workflow context through specialized 'prime' commands
Conversation compaction survival for long-running project continuity
0 GitHub stars
Graph-based dependency tracking to manage complex task blockers
Seamless team synchronization via standard Git repository integration

使用场景

Resuming development context after an LLM context window reset or compaction
Coordinating task progress between AI agents and human developers through shared git state
Executing multi-day software engineering projects with complex, interlocking requirements

关于

主要功能

Git-backed persistent memory for multi-session context recovery
AI-optimized workflow context through specialized 'prime' commands
Conversation compaction survival for long-running project continuity
0 GitHub stars
Graph-based dependency tracking to manage complex task blockers
Seamless team synchronization via standard Git repository integration

使用场景

Resuming development context after an LLM context window reset or compaction
Coordinating task progress between AI agents and human developers through shared git state
Executing multi-day software engineering projects with complex, interlocking requirements