@shutovks/headroom-opencode-standalone

headroom-opencode-standalone

Headroom context compression for OpenCode — spawns the proxy, intercepts provider traffic, exposes reversible compression.

Compresses everything your AI agent reads — tool outputs, logs, files, conversation history — before it reaches the LLM. Same answers, a fraction of the tokens. Runs locally; your data stays on your machine.

Install

# 1. Headroom CLI (one-time)
uv tool install "headroom-ai[all]"
# or: pip install "headroom-ai[all]"

# 2. The plugin
npm install @shutovks/headroom-opencode-standalone

OpenCode config (opencode.jsonc):

{
  "plugins": {
    "@shutovks/headroom-opencode-standalone": {}
  }
}

The plugin starts the local headroom proxy automatically and stops it when OpenCode exits. If no headroom binary is on PATH, it loads as a no-op — your session runs normally, just without compression.

Environment

License

MIT