The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides installation instructions (curl -sS https://webi.sh/gh | sh) that download and execute code from an external server directly in the system shell. This pattern allows for the execution of unverified remote code from a third-party source without integrity checks or local auditing.
[EXTERNAL_DOWNLOADS]: The skill references an installer from webi.sh, which is an external entity not identified as a trusted vendor or official distribution channel for the GitHub CLI.
[COMMAND_EXECUTION]: The skill uses the GitHub CLI (gh) to execute shell-based operations including managing CI/CD runs, viewing logs, and downloading files, which involve interactions with the network and local operating system.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of raw logs from GitHub Actions via the gh run view --log command. Because logs can contain output from any process running in the CI environment, they represent an attack surface where a malicious actor could embed instructions designed to influence the behavior of the agent reading the log.
Ingestion points: GitHub Action logs retrieved via gh run view as referenced in SKILL.md.
Boundary markers: Absent; log output is ingested as raw text.
Capability inventory: File system writing via gh run download and workflow control via gh run rerun and gh workflow run.
Sanitization: Absent; the skill does not implement any filtering or escaping for content retrieved from logs.

workflow