The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides an extract tool that retrieves content from arbitrary external URLs. This data is converted to Markdown and fed into the agent's context, creating an attack surface for adversarial instructions embedded in external web pages.
Ingestion points: The url parameter of the extract tool in SKILL.md allows the agent to fetch content from any web address.
Boundary markers: Absent. The documentation does not specify the use of delimiters or instructions for the agent to ignore embedded prompts within the extracted content.
Capability inventory: The skill allows the agent to perform network requests via the search and extract tools.
Sanitization: There is no evidence of sanitization or filtering to remove malicious instructions from the fetched HTML/Markdown content.
[EXTERNAL_DOWNLOADS]: The installation instructions recommend using npx to download and execute utility packages from the NPM registry at runtime.
Evidence: The stdio and SSE transport sections in SKILL.md utilize npx -y mcp-remote and npx -y supergateway to facilitate connection to the API.
[COMMAND_EXECUTION]: The skill requires manual execution of shell commands for proxy setup and environment variable configuration.
Evidence: Use of npx, export, and .env file creation as described in the Installation and API Key Configuration sections.

anysearch-mcp-server