The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality of processing external article text.
Ingestion points: The agent reads article content from provided file paths or direct text input during 'Step 2: Analyze' (SKILL.md).
Boundary markers: The workflow does not specify the use of delimiters or 'ignore' instructions to isolate the ingested article content from the agent's internal logic.
Capability inventory: The skill has access to 'bash' for file system checks and 'write' for generating prompt files and outlines.
Sanitization: There is no evidence of sanitization or filtering of the ingested text before it is analyzed to determine illustration positions and content.
[COMMAND_EXECUTION]: The skill utilizes the 'bash' tool for environment and file system verification.
Evidence: Uses test -f commands to check for the existence of the EXTEND.md configuration file in both the project directory and the user's home directory ($HOME/.baoyu-skills/) to load preferences (SKILL.md, references/workflow.md).
[EXTERNAL_DOWNLOADS]: The skill documentation refers to external package execution for its runtime environment.
Evidence: The 'OpenClaw Preflight Checks' section in SKILL.md mentions the use of npx -y bun to ensure the Bun runtime is available. While Bun is a well-known and trusted tool, npx -y downloads and executes packages from the npm registry.

baoyu-article-illustrator