The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, creating a potential surface for indirect prompt injection. * Ingestion points: Pull Request lists (gh pr list), issue lists (gh issue list), and web search results (WebSearch) are processed in the SKILL.md. * Boundary markers: The instructions do not specify the use of delimiters or clear separation between system instructions and untrusted data in the generated brief or status page. * Capability inventory: The skill can write files (docs/status.md), send notifications (./notify), and send emails via the Resend API (curl). * Sanitization: No explicit content sanitization or validation is defined for the external data before it is formatted into notifications or the public status page.
[COMMAND_EXECUTION]: The skill utilizes command-line tools for operational tasks. * Evidence: Execution of GitHub CLI (gh) and curl for interacting with external APIs as described in the Shared setup and Priority brief sections. * Security: The skill incorporates best practices by using the ./secretcurl wrapper and placeholder tokens to manage sensitive API credentials securely.

heartbeat