The Agent Skills Directory

Indirect Prompt Injection Surface: The skill ingests text from issue threads and pull requests, which creates an inherent exposure to indirect prompt injection. 1. Ingestion points: Data is fetched from GitHub using the gh api tool. 2. Boundary markers: Instructions explicitly separate data from instructions, advising the agent to treat external content as raw text. 3. Capability inventory: The skill uses gh api and python3 for data retrieval and mathematical calculations. 4. Sanitization: A dedicated logic gate is defined to detect and exclude items that contain imperative phrases designed to influence the agent's behavior.
Command Execution: The skill runs local commands for environment setup and data processing. 1. A pre-flight check executes a local Python module (setup_preflight) to verify project settings. 2. Standard GitHub CLI (gh) commands are used for data collection. 3. Inline Python scripts are utilized for statistical calculations like the Gini coefficient, which are consistent with the skill's documented purpose.

sentiment