The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from project-level files which could be manipulated to bypass guidelines or influence agent behavior.- Ingestion points: The skill reads README.md, package.json, CHANGELOG.md, and project plans using Read, Grep, and Glob tools as part of its pre-review audit and journey tracing.- Boundary markers: There are no instructions defining delimiters or warnings to the agent to disregard instructions embedded within these external documents.- Capability inventory: The skill has access to Bash, Edit, and WebSearch tools, providing a significant capability surface if an injection occurs.- Sanitization: Content read from project files is processed directly without escaping or validation before being used to shape the DX review.- [COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform environment discovery and project metadata analysis.- Evidence: The skill executes shell commands like git log --oneline -15, git diff, gh pr view, and gh repo view to determine the review context and analyze recent changes.- Context: These operations are used to identify target branches and analyze the existing developer-facing surface area of the project.

plan-devex-review