h

headroom

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Li

open-sourcetool-integration
74.2k
Stars
+6181
Stars/month
1163
Commits (90d)
10
Releases (6m)

Star Growth

+18.5k (33.3%)estimated from velocity
54.5k65.1k75.7kJul 2Sep 30

Overview

Headroom is a library, proxy, and MCP server that compresses agent inputs like tool outputs, logs, RAG chunks, and files to reduce token usage. It runs locally, offers multiple integration methods (library, proxy, agent wrap), and maintains reversible compression with local caching. It supports cross-agent memory and can learn from failed sessions.

Deep Analysis

Key Differentiator

Provides lossless compression for diverse agent inputs (JSON, code, text) with local execution and reversible retrieval.

⚡ Capabilities

  • • Token reduction via compression
  • • Local processing (no data sent externally)
  • • Multiple integration modes (library, proxy, wrap, MCP)
  • • Cross-agent memory with dedup
  • • Reversible compression with local retrieval
  • • Output token reduction
  • • Learning from failed sessions

🔗 Integrations

Claude CodeCursorCodexLangChainAgnoStrandsMCP clientsVarious coding agents and IDEs

✓ Best For

  • ✓ Reducing LLM token costs in agent workflows
  • ✓ Optimizing tool output, log, and RAG chunk ingestion
  • ✓ Developers building or running AI agents

✗ Not Ideal For

  • ✗ End-user AI applications like chatbots or image generators
  • ✗ Non-AI products

⚠ Known Limitations

  • ⚠ Requires local deployment/integration
  • ⚠ Primarily focused on input optimization

Alternatives

See all 8 headroom alternatives →

Works with headroom

Tools that integrate with headroom, often used together in the same stack.

Compare headroom

Maintain headroom?

Show your live rank in your README, or put headroom in front of every visitor to AgentoolRank.