6 projects
relapse
Self-correcting LLM call wrapper: compress input, call the model, ground-check the answer, and on hallucination restructure the prompt and retry. The runtime guard rails for RAG.
staleness
Deterministic freshness and drift linter for RAG knowledge bases: catch stale chunks, missing/future timestamps, age-threshold breaches, temporal clustering, and embed-model age before they degrade your retrieval.
redoubt
Static prompt-injection scanner for RAG corpora: catches jailbreak signatures, encoded payloads, hidden instructions, and role-play inducements before they reach the LLM.
corroborate
Deterministic answer-grounding check for LLM/RAG outputs: catch hallucinated numbers, dates, and quotes without spending a cent on LLM-as-judge.
chaffer
Lint for RAG corpora and retrievers: catch silent bugs (duplicate chunks, embed-model truncation, eval-set leakage, dim mismatch) before you ship.
dash-mlguard
Lint for ML training pipelines: catch silent bugs (leakage, drift, schema mismatch) before they ruin your model.
