The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes the GitHub CLI to search for existing issues to prevent duplicates.
Evidence: Executes gh issue list --search "<keywords>" --state all --limit 20 and gh search issues in the shell.
[PROMPT_INJECTION]: The skill processes data from external sources (existing GitHub issues) and local repository files which could theoretically contain malicious instructions.
Ingestion points: Repository issue templates located in .github/ISSUE_TEMPLATE/ and search results returned by the gh command.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious content within templates or existing issues.
Capability inventory: The agent has access to the create_issue tool and shell execution for gh commands.
Sanitization: No specific filtering or validation of the ingested external content is mentioned.

issue