llamaperf

ojuschugh1/sqz

Runs locallyNo key needed

Context compression for AI coding agents: compresses tool output before it enters the model, dedups repeats to 13-token refs. Claude Code, Cursor, Codex, Kiro, Zed, any MCP client. Rust, zero LLM calls.

View on GitHub

Stars
642
Gained, 7 days
+8
Gained, 30 days
n/a
Reddit posts
1

Rust · NOASSERTION · last push 2026-09-24 · created 2026-04-12

GitHub stars, last 90 days

2026-09-26: 6312026-10-07: 642

Does it work with local models?

Reports from people who ran it with a model on their own hardware. Tool calling depends on the model as much as on the server, so say which model and quant you used.

No reports yet.

Add your report

Sign in to say whether it worked for you.

Reddit posts linking it

In the official MCP registry

  • io.github.ojuschugh1/sqzv1.9.0Runs locallyNo key needed

    Pre-injection context compression for coding agents. Zero LLM calls, zero telemetry, offline-safe.

    Packages: cargo sqz-mcp (stdio)