The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and acts upon user-supplied 'ladder chunks' and 'design-plan.md' content which is interpolated into the agent's context.
Ingestion points: Untrusted data enters the agent context through pasted build chunks and design-plan files as specified in the brief tiers in SKILL.md.
Boundary markers: The skill employs a structured 'commit ritual' and specific workflow headers to organize its output, but lacks explicit instructions to the agent to ignore or escape instructions embedded within user input.
Capability inventory: The skill has permissions to write code to the local file system and execute verification scripts like preflight_scan.py located in the skill's root directory.
Sanitization: The instructions do not define any specific sanitization or validation protocols for the external content before it is processed by the agent.

frontend-dev