LLM claims validated against real data — never invented. Kremis ingests EAV signals, builds a deterministic graph, classifies every response: FACT (direct edge), INFERENCE (derived path), or UNKNOWN (not in data). No confidence scores. No probabilistic gaps. Built in pure Rust (no async, no floats in core), with ACID persistence via redb, HTTP API, CLI, and MCP bridge for Claude/Cursor. Apache 2.0. v0.17.8 — alpha, functional, tested. 354 tests passing.

EOL Dataset by HeroDevs — Find every EOL dependency in your stack. Free. In 5 minutes.

Find every EOL dependency in your stack. Free. In 5 minutes.

I built Kremis because RAG retrieves the right documents but the LLM still invents details. I needed a way to check each claim against real data — not with similarity scores, but with a binary answer: is this in my data or not? The core loop: 1. Ingest signals as EAV triples (entity, attribute, value) 2. Kremis builds a weighted graph from co-occurrence 3. Query → every result tagged as FACT, INFERENCE, or UNKNOWN No embeddings in core. No floating-point. Same input, same output. What's working today: - kremis-core (pure Rust, no async, no network) - HTTP API (axum) + CLI (clap) - MCP bridge — 9 tools, works in Claude Desktop and Cursor - ACID persistence via redb - Docker image The main friction is EAV ingestion: you structure your data as signals before querying. Is that too rigid? I'd like to hear. Demo in the repo (Python, stdlib only): python examples/demo_honesty.py Docs: https://kremis.mintlify.app Background story: https://dev.to/tykolt/i-spent-mo...

EOL Dataset by HeroDevs — Find every EOL dependency in your stack. Free. In 5 minutes.

Find every EOL dependency in your stack. Free. In 5 minutes.

kremis

Every LLM claim traced to real data — or rejected

Every LLM claim traced to real data — or rejected