The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to investigate the repository and perform actions, specifically gh issue create for creating GitHub issues and git log for checking file history.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes untrusted data from the repository's source files and git logs to generate its output.
Ingestion points: The agent reads source files, dependency lists, and git logs during the "Explore and diagnose" phase (SKILL.md).
Boundary markers: There are no explicit boundary markers or instructions to the agent to disregard embedded instructions within the codebase files it reads.
Capability inventory: The skill possesses the ability to write to the local file system (creating markdown files in ./issues/) and the ability to create external records via the GitHub CLI.
Sanitization: No sanitization or verification of the extracted content is performed before passing it to the issue creation command.
[PROMPT_INJECTION]: The skill includes instructions that reduce human oversight, explicitly directing the agent to "minimize questions to the user" and to "NOT ask the user to review before creating" the GitHub issue or local file. This autonomy increases the risk that an indirect injection could trigger automated, unintended external actions.

triage-issue