The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes project requirements and source code during the implementation and review phases, which constitutes an ingestion surface for potentially malicious instructions embedded in data.
Ingestion points: Source code and requirement documents processed by /codex-implement and /codex-review-fast.
Boundary markers: No explicit delimiters or instructions are provided to the agent to disregard instructions contained within the code or requirements files.
Capability inventory: The skill has access to powerful tools including Bash, Write, and Edit, which could be abused if the agent follows instructions embedded in source data.
Sanitization: There is no mention of input validation or sanitization for the code or requirements being handled.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute verification and precommit scripts (/verify, /precommit). While these are standard development tasks, they involve executing shell commands based on the project environment.
[SAFE]: The skill explicitly prohibits dangerous git operations like push and commit within its workflow, enforcing a manual gate for version control changes.

feature-dev