discovery-catalog
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Repository Cataloging Logic: The skill uses a PowerShell script to inventory the repository's contents by reading metadata files such as metadata.yaml and kit.json. This is a standard functional component for discovery and organization within the repository environment.
- Read-Only Capability: All operations performed by the skill are read-only, focusing on file system traversal and data display without modifying the environment, making network calls, or accessing sensitive credentials.
- Indirect Prompt Injection Surface: The skill processes and displays content from files within the repository, which constitutes a potential consideration for indirect prompt injection.
- Ingestion points: External data enters the agent context through the reading of metadata.yaml, kit.json, and tool.yaml files in the agents/ and starter-kits/ directories via the scripts/discovery-catalog.ps1 script.
- Boundary markers: The skill is instructed to format its output as Markdown tables, which serves as a structural delimiter for the processed data.
- Capability inventory: The skill is authorized to use the shell tool for executing the inventory script, which performs directory listings and file reads.
- Sanitization: The script performs basic sanitization by escaping pipe characters (|) in descriptions to maintain the integrity of the Markdown table output.
Audit Metadata