The Agent Skills Directory

[SAFE]: The skill provides instructions for deploying Firecrawl using official resources and secure patterns.
[CREDENTIALS_UNSAFE]: The skill follows security best practices by instructing users to use platform-specific secret managers (e.g., gcloud secrets, vercel env) rather than hardcoding sensitive API keys in the deployment scripts. The use of ${BULL_AUTH_KEY:-changeme} in the Docker configuration is a standard placeholder for environment-based configuration.
[EXTERNAL_DOWNLOADS]: The skill references official Docker images for Firecrawl (mendableai/firecrawl) and Redis (redis:7-alpine). These are well-known, trusted sources for the intended integration.
[INDIRECT_PROMPT_INJECTION]: The skill describes an integration that processes external data from web scraping and webhooks, representing a standard attack surface for this type of application. It includes mitigations such as HMAC-based signature verification for webhooks and recommendations to avoid leaking scraped content in logs.
Ingestion points: app/api/scrape/route.ts (API route ingesting external URLs) and app/api/webhooks/firecrawl/route.ts (Webhook listener ingesting external data).
Boundary markers: Implements x-firecrawl-signature verification using SHA256 HMAC.
Capability inventory: The skill is configured with restricted shell access (Bash) for standard deployment tools (vercel, docker, gcloud).
Sanitization: Recommends using generic error messages and preventing data leakage in logs.

firecrawl-deploy-integration