The Agent Skills Directory

[COMMAND_EXECUTION]: The skill leverages the GitHub CLI (gh) to interact with issues and Project v2 boards.
It instructs the agent to request the highly privileged admin:org scope via gh auth refresh to perform administrative tasks like creating issue types at the organization level.
It uses gh api and gh api graphql to perform state-changing operations (PATCH, POST, Mutations) on GitHub repository items.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) through the processing of external repository data.
Ingestion points: Data enters the agent context from GitHub issue titles, bodies, and project fields through commands like gh project item-list and the gh seq tool.
Boundary markers: There are no explicit delimiters or safety instructions (e.g., "ignore embedded commands") used when the agent processes data from these external sources.
Capability inventory: The skill allows the agent to create new issues, modify existing issue types, and update project metadata across multiple repositories.
Sanitization: No sanitization or validation of ingested issue content is performed before it is used to influence the agent's task-sequencing or update logic.

atlas-crew-tasks