The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources (GitHub and Linear issue trackers) which could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: The skill fetches issue identifiers, URLs, titles, states, and relationships from the tracker providers as specified in the 'Resolve the tracker contract' section.
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the fetched issue content as untrusted text or to ignore embedded instructions within issue titles or descriptions.
Capability inventory: The skill utilizes gh issue develop (GitHub CLI), git operations (fetch, checkout, worktree management), and has the ability to mutate tracker states (status transitions and assignment).
Sanitization: There is no evidence of sanitization or escaping of the issue metadata before it is used to rename tasks or determine the engineering context.
[COMMAND_EXECUTION]: The skill performs several local shell commands to manage the engineering environment. These are consistent with the skill's stated purpose of aligning git worktrees and branches with issue trackers.
Evidence: Use of gh issue develop, git worktree list --porcelain, and various git fetch and git checkout commands.

working-on-issues