The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the execution of a local shell script to handle asset downloads.
Evidence: SKILL.md instructs the agent to execute bash scripts/fetch-stitch.sh to download HTML code from Stitch design URLs.
[EXTERNAL_DOWNLOADS]: The skill fetches external content from the Stitch service during execution.
Evidence: The provided script scripts/fetch-stitch.sh utilizes curl to download design assets from remote URLs obtained via the Stitch MCP tools.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external designs which may contain malicious instructions designed to influence the agent's code generation.
Ingestion points: As described in SKILL.md, the skill retrieves design JSON and HTML code from the Stitch MCP service (e.g., via get_screen).
Boundary markers: There are no explicit instructions or delimiters defined to ensure the agent ignores or sanitizes natural language instructions that might be embedded within the retrieved design data.
Capability inventory: The agent has the capability to execute shell commands, perform network downloads, and write code files to the project directory based on the fetched content.
Sanitization: The skill does not define any validation or sanitization steps for the data retrieved from external design files before processing it for code generation.

stitch-ui-uviewpro-components