sqz runs on your own machine — the client starts it, so there is no endpoint to ping. Last commit 11 Oct 2026.
Pre-injection context compression for coding agents. Zero LLM calls, zero telemetry, offline-safe.
What the code does
We read the source, 5 d ago · rules ccca72095570
What this server is able to do. For an MCP server this is often the job itself — a terminal server runs commands because that is what it is for. Listed so you know what you are plugging in, not as an accusation.
val pb = ProcessBuilder(cmd)
private exec(args: string[], input?: string): string {
result = subprocess.run([str(binary)] + sys.argv[1:])
resolved = posixpath.normpath(posixpath.join(folder, path_only))
Is this your server and something here is wrong? Tell us — corrections are free and do not require a plan.
This code can reach further than it looks
We found places where it runs commands, builds paths or queries from values it is given. None of that is a flaw by itself — it becomes one when the code changes, and code changes quietly between releases. We re-read it on every one.
Alternatives to sqz
same job, measured the same wayPrompt-injection firewall for AI agents — scan untrusted text before LLM calls.
Offline persistent memory for AI coding agents — 35 tools, 80% fewer calls, no LLM
Context selection and compression for AI agents - recall measured, not claimed.
Checkpoints, timeline and rewind for AI coding agents. Local, no telemetry.
Deterministic context selection for AI coding agents. Local memory that compounds across sessions.
Scan skills, MCP configs, and agent content for injection and supply-chain risks. Offline, no LLM.
Project memory for coding agents: requirements, decisions, code graph and delivery telemetry.
Verified context layer for AI coding agents — prunes, compresses, indexes, remembers. 32 MCP tools.
sqz — questions
Answers built from our own checks of this server.