The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill is highly vulnerable to indirect prompt injection. 1. Ingestion points: The scripts test_suite_generator.py and coverage_analyzer.py ingest untrusted project source code. 2. Boundary markers: None are present in the skill documentation to delimit data from instructions. 3. Capability inventory: The skill has high-privilege capabilities including kubectl apply, docker build, and automated file writing ('fixes'). 4. Sanitization: No sanitization of ingested code content is mentioned.
[Unverifiable Dependencies & Remote Code Execution] (MEDIUM): The setup instructions utilize npm install and pip install to fetch dependencies from public registries without integrity checks or version pinning in the main file, posing a supply chain risk.
[Dynamic Execution] (MEDIUM): The skill documentation describes 'automated fixes' generated by the coverage analyzer, implying the modification or generation of executable code at runtime based on analyzed content.

senior-qa