Sign in to view source links and access this dataset
Description
AI-Secure released a collection of agent trajectories from the DecodingTrust-Agent Platform. The data likely contains logs from evaluating AI agents across 14 real-world domains and over 50 simulation environments replicating systems like Google Workspace, PayPal, and Slack. The dataset was last updated on June 1, 2026.
Use Cases
Analyze agent failure modes based on interactions with simulated real-world systems.
Benchmark agent robustness based on performance across 14 different domains.
Train or fine-tune AI agents based on recorded trajectory data.
Study agent behavior in simulated environments replicating platforms like Salesforce and Databricks.
Strengths
Covers 14 distinct real-world domains for evaluation breadth.
Includes simulations of widely-used systems such as Google Workspace, PayPal, and Slack.
Contains trajectories from over 50 different simulation environments.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
Source
AI-Secure via the DecodingTrust-Agent Platform.
Collection Method
Produced from evaluating AI agents in controlled simulation environments.
Freshness
Last updated 2026-06-01 23:06:20; freshness should be verified.
License is unknown; terms of use must be verified before application.