Context Aware Repository Prompt Injection: Training Data for AI Security Scanners

ShieldX created this dataset for training AI security scanners to detect repository-aware prompt injection attacks. The dataset focuses on malicious instructions embedded in code repositories, documentation, comments, and other project artifacts. It was last updated on June 8, 2026.

Use Cases

• Train AI security scanners based on examples of malicious instructions in code repositories.
• Evaluate detection models for prompt injection attacks based on patterns in documentation and comments.
• Benchmark AI system robustness against adversarial inputs based on examples from configuration files and issue trackers.

Strengths

• Dataset is specifically designed for training and evaluating AI security scanners.
• Focuses on repository-aware prompt injections, a targeted attack vector.
• Created by ShieldX, an organization specializing in AI security.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect bias inherent to the specific examples curated by the author.

Provenance

Source

ShieldX

Freshness

Last updated 2026-06-08 12:20:17.