The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands, specifically git diff, git log, and git rev-parse, using a user-supplied <fixed-point> argument. While the skill includes a validation step using rev-parse, the direct interpolation of user input into shell commands presents a risk of command injection if the underlying agent platform does not provide automated escaping.
[PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection by processing external data and passing it to sub-agents.
Ingestion points: Data is ingested from git commit messages, the docs/agents/issue-tracker.md file, and various specification files found in docs/, specs/, or .scratch/ directories.
Boundary markers: The prompts provided for the 'Standards' and 'Spec' sub-agents do not utilize explicit boundary markers (like XML tags or unique delimiters) or safety instructions to prevent the sub-agents from following malicious commands embedded in the commit messages or specs.
Capability inventory: The skill has the capability to execute shell commands, read the file system, and spawn new agent instances.
Sanitization: There is no documented logic for sanitizing or escaping the content retrieved from external sources before it is interpolated into the prompts for the sub-agents.

code-review