The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches the primary application source code from a public GitHub repository.
Evidence: git clone https://github.com/fahad-hamid/psique-workflow-clinic.git in SKILL.md.
[COMMAND_EXECUTION]: The setup process requires executing shell commands for package installation and database management.
Evidence: npm install, npx prisma generate, and npx prisma migrate deploy commands in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The application ingests external, untrusted data via WhatsApp messages which are subsequently processed for automated responses and crisis detection alerts.
Ingestion points: WhatsAppService.handleIncomingMessage reads content from messages.upsert events.
Boundary markers: None identified in the provided snippets for the message handling logic.
Capability inventory: The skill has capabilities to send messages (socket.sendMessage), perform network API calls (Anthropic, LiveKit, AFIP), and query the database via Prisma.
Sanitization: Basic keyword matching (.toLowerCase().includes()) is used to detect crisis terms but does not sanitize content against prompt injection if the data is passed to the AI service.

sesion-mental-health-clinic-platform