The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses strong instructional language such as 'IMPORTANT', 'MANDATORY', and 'MUST ATTENTION' to enforce task planning and quality control. These instructions do not attempt to bypass safety guidelines or override system prompts; they are operational best practices for the agent's persona.
[DATA_EXFILTRATION]: No network operations (such as curl, wget, or fetch) or hardcoded credentials were identified. The skill interacts exclusively with local project files in directories like 'docs/' and 'team-artifacts/'.
[COMMAND_EXECUTION]: The skill references 'Glob' for file discovery and 'TaskCreate' for workflow management. These are used within the scope of navigating documentation and organizing sub-tasks, posing no risk of arbitrary system command execution.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted documentation data from 'docs/business-features/'. While this is an ingestion surface, the skill implements boundary markers and manual verification steps ('AskUserQuestion' validation interview) to mitigate risks. The capabilities are limited to generating structured project artifacts (user stories, PBIs).
[SAFE]: The skill adheres to the 'AI Mistake Prevention' guidelines and 'Critical Thinking Mindset', which are focused on accuracy, source citation, and logical consistency rather than malicious intent.

business-analyst