Automated Failure Attribution in Long-Horizon Agent Runs
A research project that answers the question every failed agent run raises: which step broke it? Building an attributed corpus...
Tagged: trajectory-analysis
2 ideas with this tag
Trajectory-Anomaly Pre-Commit Detection of Multi-Step Injection Attacks
A research project that trains unsupervised models on benign agent trajectories to detect multi-step prompt-injection campaigns before the harmful action...