The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local TypeScript script using npx ts-node as defined in the metadata and provided examples.
[EXTERNAL_DOWNLOADS]: The scripts/fetch.ts file performs network requests to external URLs provided via command-line arguments, using the coze-coding-dev-sdk to retrieve web pages, documents, and images.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) by processing untrusted data from the internet.
Ingestion points: The fetchUrl function in scripts/fetch.ts retrieves content from external URLs.
Boundary markers: The formatting functions (formatAsText and formatAsMarkdown) use simple text headers (e.g., FETCHED CONTENT, ---) but do not include explicit instructions for the agent to ignore potential commands embedded within the retrieved data.
Capability inventory: The agent using this skill may have access to sensitive tools or file system operations that could be targeted by malicious instructions found in the fetched content.
Sanitization: No filtering or validation of the fetched content is performed before it is presented to the agent.

coze-web-fetch