The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies surfaces where the agent ingests and analyzes untrusted external data, which could contain malicious instructions.
Ingestion points: The agent is instructed to read error messages, stack traces, and user-provided issue descriptions in SKILL.md and INCIDENT-RESPONSE.md.
Boundary markers: The instructions do not specify explicit delimiters or isolation techniques for the untrusted data being analyzed.
Capability inventory: The skill leverages other capabilities like the debug skill for root cause analysis and deploy for hotfixes or rollbacks.
Sanitization: The instructions include proactive warnings to avoid logging sensitive data such as passwords, API keys, or session tokens.
[EXTERNAL_DOWNLOADS]: The skill guides the user to install well-known error tracking packages and utilize established monitoring services.
Evidence: Recommends installing the @sentry/nextjs package and configuring integrations with services such as UptimeRobot, Stripe, and SendGrid.
[COMMAND_EXECUTION]: The documentation provides templates for executing standard shell-based version control operations for deployment recovery.
Evidence: INCIDENT-RESPONSE.md outlines procedures using git revert and git push to manage production rollbacks.

monitor