The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell execution to run local scripts such as init-room.sh, status.sh, and build-html.sh. These scripts are resolved from plugin installation directories (e.g., ~/.claude/plugins/cache/) and are used to initialize the workspace and compile the final HTML deck. This behavior is standard for the tool's documented purpose.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and audits user-supplied and AI-generated content (e.g., slides.html, custom.css). This creates a surface where instructions embedded in the slide content could potentially influence the agent's review personas or narrative logic.
Ingestion points: Reads files such as slides.html, custom.css, deck.md, DESIGN.md, and MEMORY.md into the agent's context.
Boundary markers: The instructions lack explicit boundary markers or directives to ignore instructions embedded within the slide files during the review process.
Capability inventory: The skill possesses file-writing capabilities and can execute local shell scripts to build project artifacts.
Sanitization: The skill does not perform content sanitization on the audited HTML or CSS files before processing them for visual and quality feedback.

codeck-review