The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform local environment checks using shell and PowerShell commands to detect configuration files. Evidence: Step 1.1 in SKILL.md defines the use of test -f and Test-Path to check for EXTEND.md in the project and user home directories.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted user content and incorporates it into a prompt for an external image generation tool. Ingestion points: Source content provided as a file path or pasted text in SKILL.md (Step 1.2). Boundary markers: The skill uses the {{CONTENT}} placeholder in references/base-prompt.md to delineate user data. Capability inventory: The skill generates a prompt and triggers an external image generation skill in Step 6. Sanitization: While the skill mandates stripping credentials and secrets, it lacks specific sanitization or escaping mechanisms to prevent adversarial instructions within the user data from influencing the final prompt.

baoyu-infographic