The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by design, as it must ingest and obey instructions from external project documents.
Ingestion points: Reads UI decisions and requirements from plan.md, PRD, DESIGN.md, and contract.json (SKILL.md).
Boundary markers: Absent. The skill does not specify any delimiters or safety instructions to prevent the agent from following malicious commands embedded in these input files.
Capability inventory: The skill has the ability to execute a local script (scripts/visual-state.mjs) and write to files in the .helloagents/ directory (SKILL.md).
Sanitization: Absent. There is no evidence of validation or filtering of the content retrieved from external sources before processing.
[COMMAND_EXECUTION]: The skill performs shell command execution to handle visual verification state management.
Evidence: Instructs the agent to call scripts/visual-state.mjs write to record check results in .helloagents/.ralph-visual.json (SKILL.md).

hello-ui