The Agent Skills Directory

[PROMPT_INJECTION]: No instructions were found that attempt to override the AI's core behavior, disable safety filters, or extract system prompts. The instructions are focused on maintaining coaching discipline.
[DATA_EXFILTRATION]: There are no indicators of data exfiltration. The skill does not use network tools like curl or wget to send information to external servers. It manages user context using local templates and explicitly instructs the agent not to record sensitive data without consent.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process personal narratives from users. It manages the associated risks through a well-defined structure:
Ingestion points: User-provided descriptions of life choices, emotions, and goals (defined in SKILL.md).
Boundary markers: The skill includes high-priority "Safety Boundary" instructions to stop the coaching process if crisis signals are detected and requires explicit consent before updating the user profile.
Capability inventory: The skill writes to local files in the templates/ directory to maintain state.
Sanitization: The skill uses structured templates (USER_CONTEXT.md and SESSION_STATE.md) to organize the agent's interpretation of user input.
[SAFE]: The external references provided for frameworks (e.g., APA, ICF, and academic institutions) are from reputable and authoritative domains. The skill follows best practices for coaching ethics and professional boundaries.

life-coach