The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that explicitly direct the agent to suppress human oversight and bypass secondary approval gates for repository and tracker mutations. The instructions state: 'Treat an invocation whose scope requests this workflow as authorization for its stated branch and tracker mutations... do not add a second approval gate when the request already specifies the workflow.' This increases the risk of the agent performing destructive or unintended actions without user review.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could contain embedded instructions to influence the agent's behavior.
Ingestion points: Specifications and tickets loaded from external issue trackers or markdown files.
Boundary markers: None. The instructions do not provide delimiters or warnings to ignore instructions embedded within the spec or ticket data.
Capability inventory: The skill possesses significant capabilities, including branch creation, code merging, pull request management, and issue tracker modification (closing/resolving tickets).
Sanitization: None. There is no evidence of filtering or sanitizing the input data before it is used to guide the implementation or merge process.
[COMMAND_EXECUTION]: The skill relies on shell-level operations for repository management and tool invocation.
Evidence: The instructions require the agent to perform git branch creation, worktree management, and merging, as well as calling tools like tdd and code-review via the Skill tool interface.

implement-spec