The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted writing samples and legacy files to create durable instructions (VOICE.md, STYLE.md) that guide the agent's future behavior. This creates a surface for indirect prompt injection where malicious content in a writing sample could influence the generated guidelines. * Ingestion points: User-provided writing samples and files like TASTE.md are used to calibrate voice and style profiles. * Boundary markers: The skill does not define specific delimiters or instructions to isolate user-provided content from the generated guidelines. * Capability inventory: The skill has the ability to read and write files in the local workspace and execute project setup commands. * Sanitization: No explicit sanitization or instructions to ignore embedded commands within user samples are present.
[COMMAND_EXECUTION]: The skill instructions include running cw-setup-project to initialize the workspace directory structure. This appears to be a local setup tool specific to the vendor's workflow.

cw-onboarding