The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes content from GitHub issues, which are external and user-controlled.
Ingestion points: In VIEW_ISSUE.md and LIST_ISSUES.md, the agent is instructed to fetch issue details including titles, bodies, and comments.
Boundary markers: No specific boundary markers or instructions are provided to the agent to distinguish between its instructions and the data fetched from GitHub.
Capability inventory: The skill possesses the capability to write to the repository (create/edit/comment/close/assign) and execute gh CLI commands.
Sanitization: There is no evidence of sanitization or safety-filtering for the content retrieved from external sources.
[DYNAMIC_EXECUTION]: The skill employs dynamic script generation and execution techniques at runtime.
Evidence: CREATE_ISSUE.md includes a line that executes node to programmatically determine the temporary directory and a timestamp for naming files.
Evidence: UPDATE_ISSUE.md and ASSIGN_ISSUE.md provide instructions to use shell heredocs (cat << 'EOF') to create temporary markdown files on the fly.
[COMMAND_EXECUTION]: The skill operates by executing shell commands via the gh CLI, using arguments that may be derived from potentially untrusted external issue data.

github