The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash and Agent tools to inspect local files and project architecture. This is used to provide grounded and specific risk analysis as per the skill's instructions. No evidence of malicious command construction was found.
[PROMPT_INJECTION]: The skill's instructions are focused on structured failure analysis. It does not contain patterns meant to bypass AI safety guardrails or extract system prompts.
[DATA_EXFILTRATION]: There are no network requests or data transmission patterns in the skill's logic. It operates on local context for analysis purposes.

pre-mortem