The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by ingesting untrusted data from external sources.
Ingestion points: The skill reads the title and body of GitHub issues using the gh issue view command in Step 1 of SKILL.md.
Boundary markers: There are no clear delimiters or instructions to the agent to ignore potentially malicious instructions embedded within the issue body.
Capability inventory: The skill possesses high-privilege capabilities including branch creation (git checkout -b), code pushing (git push), pull request creation (gh pr create), and project board modification (gh project item-edit).
Sanitization: The untrusted issue content is not sanitized or validated before being presented to the 'Product Council' and 'Feature Council' personas for evaluation.
[COMMAND_EXECUTION]: The skill utilizes multiple system-level CLI tools to perform its tasks.
Evidence: Uses git for repository management, gh for GitHub API interactions, and pnpm for running the Prettier formatter.
Context: These operations are core to the skill's functionality for planning and documenting features.
[COMMAND_EXECUTION]: The skill generates and executes short Python scripts at runtime to process data.
Evidence: In SKILL.md Steps 7 and 9, the skill uses python3 -c to parse JSON output from the GitHub CLI to extract specific metadata like item IDs.

plan-feature