The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines procedures for the agent to execute system commands such as npm and git for test automation and impact analysis.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the processing of external, untrusted data. * Ingestion points: Input includes changed_files and production_bugs descriptions from potentially external sources. * Boundary markers: There are no explicit markers or instructions to isolate or ignore instructions embedded within the processed data. * Capability inventory: The agent is instructed to use subprocess capabilities (npm, git) which could be targeted by malicious input. * Sanitization: The skill does not define methods for sanitizing or validating external textual data before processing.

regression-testing