Main repository (flagship). Reproducible research tool for collecting, qualifying and analyzing web corpora.
-
Updated
Jul 4, 2026 - Python
Main repository (flagship). Reproducible research tool for collecting, qualifying and analyzing web corpora.
git diff for prompt engineers
RE:GE — A symbolic operating system for myth, identity, ritual, and recursive systems. 21 organ handlers, ritual syntax DSL, workflow orchestration, external bridges (Obsidian/Git/Max-MSP).
Drop-in safety layer that stops semantic caches from silently returning wrong answers — with the first independent open benchmark of semantic-cache false-hit rates. Free, local, LiteLLM + OpenTelemetry.
An MCP that reviews your architecture, estimates real capacity from cited numbers, surfaces the trade-offs you are making, and roasts bad design in a dry voice. Deterministic engine, no model, runs free and local.
Financial AI research workspace with a Graham-style agent, evidence-grounded RAG, SEC/Stooq market data, mock/live data modes, and public-safe prompt configuration.
Turn any Git repository into a local SWE-bench-style coding-agent benchmark.
Prompt regression testing for CI with deterministic judges and no-key demos.
Security-conscious JSON parsing with duplicate-key detection and depth/size limits.
An AI-powered operating environment I built to act like a personal Jarvis. It listens for a wake word, understands voice commands, controls your browser and desktop apps, manages files, and runs multi-step workflows automatically. Local-first, extensible, and genuinely useful.
Tiny inspectable agent runtime with tools, memory, tracing, and safe no-key examples.
Add a description, image, and links to the flagship topic page so that developers can more easily learn about it.
To associate your repository with the flagship topic, visit your repo's landing page and select "manage topics."