The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, including pull requests, code diffs, and library source code, which could contain malicious instructions designed to influence the agent.
Ingestion points: The instructions in SKILL.md direct the agent to read diffs, pull requests, and external library source code.
Boundary markers: The skill does not define specific delimiters or instructions to ignore instructions embedded within the code or comments being analyzed.
Capability inventory: The skill possesses code execution capabilities, as it is instructed to write and run scripts to verify code behavior.
Sanitization: No specific sanitization or filtering logic is prescribed for the data ingested from external sources.
[DYNAMIC_EXECUTION]: The skill's core functionality relies on generating and executing scripts at runtime to validate the behavior of the software being analyzed.
Execution pattern: The instructions explicitly direct the agent to "Write a script or test that runs the real code, run it, and paste what happened."
Purpose: This behavior is aligned with the skill's primary objective of performing technical impact analysis and proving safety facts through empirical testing.

blast-radius