Beads

Tracked

A lightweight framework that provides persistent memory and efficient retrieval for code agents.

Author Steve Yegge Open Sourced 2025-10-12 Last Commit Unknown

Beads is a lightweight memory layer designed for coding agents, providing persistent context and efficient retrieval to enhance AI-assisted development. It converts important conversation snippets and code context into embeddings, stores them in an efficient index, and enables reliable retrieval during multi-turn interactions or long-lived sessions.

Persistent Memory

  • Stores key conversation snippets, code fragments, and metadata across sessions
  • Survives agent restarts and context window resets
  • Decouples memory management from model inference for cleaner architecture
  • Lightweight storage footprint optimized for coding assistant workloads

Embedding-Based Retrieval

  • Vector index for fast semantic search across stored memories
  • Metadata filtering to narrow results by file, language, or topic
  • Low-latency queries tuned for real-time coding assistant interactions
  • Relevant context returned in compact form for direct injection into the prompt

Integration Design

  • Simple, extensible API for plugging into existing agent runtimes and toolchains
  • Works alongside any LLM by appending retrieved context to the model's input
  • Modular architecture that separates memory concerns from reasoning logic
  • Compatible with popular coding agent frameworks and IDE extensions

When to Use Beads

  • Coding assistants that need to maintain conversational state across sessions
  • Recovering relevant past changes, annotations, and debugging history
  • Enriching code generation and debugging with historical project context
  • Reducing context engineering complexity in long-lived agent workflows