The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses user-supplied input to construct and execute shell commands, specifically git diff <fixed-point>...HEAD, git log <fixed-point>..HEAD --oneline, and git rev-parse <fixed-point>. While the skill attempts to validate the reference with rev-parse, if the environment does not properly sanitize arguments before execution, this represents a potential command injection surface.
[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection. It ingests untrusted data from the repository's git diff and specification files (local or remote) and interpolates this content directly into the prompts for parallel sub-agents.
Ingestion points: Git diff output and various specification sources (e.g., .scratch/, docs/, specs/, or remote GitHub/GitLab issue trackers).
Boundary markers: Absent. The skill does not use delimiters or instructions to the sub-agents to ignore embedded commands within the processed data.
Capability inventory: The skill uses git and file system read capabilities; it spawns sub-agents via the Agent tool to perform the analysis.
Sanitization: No sanitization or escaping of the diff or spec content is performed before passing it to the sub-agents.

code-review