The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute a Python script located at stitch-skills/plugins/stitch-design/skills/upload-to-stitch/scripts/upload_to_stitch.py for design system uploads.
[CREDENTIALS_UNSAFE]: The execution command for the upload_to_stitch.py script instructs the agent to pass an --api-key as a command-line argument. This is a security risk as credentials passed via CLI arguments can be exposed in process monitors, logs, or command history.
[INDIRECT_PROMPT_INJECTION]: The skill fetches HTML content from external URLs using read_url_content to analyze and synthesize design systems. This untrusted data could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: External HTML content is retrieved via read_url_content in SKILL.md (Retrieval step 4).
Boundary markers: No explicit instructions or delimiters are provided to ensure the agent ignores instructions embedded within the fetched HTML.
Capability inventory: The skill has access to Bash (script execution), Write (file system access), and stitch*:* (API operations).
Sanitization: There is no evidence of sanitization, filtering, or validation of the external content before it is processed by the agent.

stitch-manage-design-system