The Agent Skills Directory

[SAFE]: The skill implements a comprehensive crisis detection system using explicit and passive keyword triggers to identify potential self-harm or psychological distress. It mandates the inclusion of crisis resources and disclaimers in all outputs, demonstrating a high degree of safety awareness for its specific domain.
[SAFE]: The skill uses internal agent capabilities (Skill, Agent) to modularize different counseling frameworks (e.g., CBT, ACT, IFS). These dependencies refer to internal skill modules rather than external, untrusted code or packages.
[SAFE]: The routing logic is clearly defined in router_prompt.md and references/pipelines.md, providing a transparent mapping of user situations to established psychological methodologies.
[SAFE]: All external URLs referenced in the skill point to well-known, official psychological support organizations and professional directories (e.g., befrienders.org, psychologytoday.com), which are appropriate for the tool's intended purpose.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided scenarios to determine the best counseling framework. 1. Ingestion points: User input is interpolated into the {scenario} variable within router_prompt.md. 2. Boundary markers: The prompt uses a markdown header (## Situation) as a delimiter. 3. Capability inventory: The skill utilizes Read, Write, Edit, Skill, and Agent tools. 4. Sanitization: There is no explicit sanitization of the user scenario.

counsel