Signal Filter
Noise cancelling for text.
Meeting transcripts, papers, and logs are mostly filler. This tool drops sentences that add no information value and returns the load-bearing lines.
pip install eyeque-signal-filter
Runs entirely in your browser
Make AI output
auditable, not just fluent.
Three local-first tools that score relevance, measure note decay, and turn LLM generations into verifiable receipts.
Three tools. One promise.
Local processing. No cloud. Every tool is fully functional in this browser.
Memory Thermometer
Find notes going stale.
Point it at a folder of Markdown and get a heat report: which notes have decayed past usefulness, ranked by recency-weighted relevance.
pip install eyeque-memory-thermometer
Flagship
Deterministic Logger
An audit trail for local LLMs.
Wraps a generation and records seed, temperature, exact prompt, model, and timestamp — then emits a SHA-256 receipt so the output can be replayed or defended later.
pip install eyeque-deterministic-logger
Browser demos use the same math as the CLIs. The web demo fingerprints your input locally; it does not call a model.
How it works
1
Paste text. Drop a transcript, log, or raw output into a tool.
2
Get a score or receipt. Signal Filter drops the noise. The Thermometer returns a vitality percentage. The Logger gives you a cryptographic fingerprint.
3
Act on it. Use the CLI to batch-process files, or the browser demos to test interactively.