The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill automatically installs a GitHub CLI extension from a third-party repository (agynio/gh-pr-review) if it is not already present. This downloads code from an external source not identified as a trusted organization.
[REMOTE_CODE_EXECUTION]: By using gh extension install, the skill facilitates the download and subsequent execution of external code. This installed code is then utilized in various workflows to manage PR threads.
[COMMAND_EXECUTION]: The skill instructs the agent to execute multiple shell commands, including complex pipelines using bash, grep, jq, and xargs. Specifically, it uses xargs to perform bulk operations based on previous tool outputs.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (PR thread IDs) and interpolates them directly into shell commands, creating an injection surface.
Ingestion points: Thread information is ingested via the gh pr-review threads list command as described in SKILL.md.
Boundary markers: There are no markers or delimiters used to isolate the thread IDs from the command structure in the bulk resolve example.
Capability inventory: The skill has the capability to install extensions, read repository metadata, and execute write operations on pull request threads.
Sanitization: There is no evidence of sanitization or validation of the thread IDs extracted by jq before they are passed to xargs and executed.

pr-threads-resolve