crashlabsai

Popular repositories Loading

1. crashedout crashedout Public

   A curated collection of real-world AI failures, mishaps, and cautionary tales. This project documents when AI systems go wrong, from chatbots giving bad legal advice to voice assistants accepting …

   TypeScript 5

2. agent-mafia agent-mafia Public

   A deterministic Mafia environment for studying AI deception, with replayable games and claims checked against logged game state.

   JavaScript 2

3. agent-check agent-check Public

   CI/CD for AI agents. Rerun your agent team in a simulated world on every PR and block merges that break behavior.

   JavaScript

4. evidencegraph evidencegraph Public

   A research prototype for investigating AI agent incidents by reconciling transcripts with independent records and producing cited findings.

   Python

5. trojan-harbor trojan-harbor Public

   An indirect prompt injection evaluation suite for Harbor, with eight matched attack and clean task families and deterministic grading.

   Python