The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute included scripts, detect-plan-context.sh and analyze-impact.sh, which automate git diffs and context gathering using the local repository environment.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing untrusted repository metadata and issue tracking information.
Ingestion points: Git branch names, commit counts, and GitHub issue titles (retrieved via the gh CLI) are processed by scripts/detect-plan-context.sh.
Capability inventory: The skill employs Bash for command execution and Task for spawning sub-agents (e.g., Explore) with dynamically generated prompts based on the ingested metadata.
Boundary markers: The skill lacks explicit protective delimiters or instructions to prevent the agent from interpreting potentially malicious instructions contained within the ingested git or issue metadata.
Sanitization: Ingested strings from the repository and external issue tracker are interpolated into report templates and agent tasks without validation or filtering.

plan-viz