The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/blockers.py script executes the gh (GitHub CLI) utility using subprocess.run to interact with the GitHub API and modify issue states. Evidence: The gh function in blockers.py wraps subprocess.run(["gh", *args], ...) and is used for operations like issue edit and api graphql.
[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it instructs the agent to interpret untrusted content from GitHub issues to make logical decisions. 1. Ingestion points: The skill fetches issue titles and bodies from GitHub repositories via the read_issue and open_issues functions in scripts/blockers.py. 2. Boundary markers: While the subagent brief provides a structured task, it lacks explicit delimiters or instructions to ignore potential commands embedded within the retrieved issue text. 3. Capability inventory: The skill has the ability to modify GitHub issue bodies and 'blocked-by' metadata through the keep and drop commands in scripts/blockers.py. 4. Sanitization: There is no evidence of sanitization or filtering of the content retrieved from issue bodies before it is processed by the agent or recorded back to GitHub.
[EXTERNAL_DOWNLOADS]: The skill fetches issue data and repository information from GitHub to perform its analysis. Evidence: Functions in scripts/blockers.py perform GraphQL queries against GitHub to retrieve issue nodes and their metadata.

settle-blockers