The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill connects to an external service provider, Unbrowse AI (unbrowse.ai), to perform searches on python.org. This is the core functionality of the skill, and the resources (URLs and MCP server endpoints) are consistent with the vendor's own infrastructure.
The skill fetches MCP configuration from https://unbrowse.ai/mcp/python.org and provides REST API access via https://unbrowse.ai/api/v1/sites/python.org/.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data by searching and rendering content from docs.python.org. This creates a potential surface for indirect prompt injection if the processed documentation contains malicious instructions.
Ingestion points: Data enters the agent context through the docs_python_org__render_page tool and various unbrowse.* tools that fetch web content (found in SKILL.md).
Boundary markers: Not explicitly defined in the skill instructions; the agent is simply told to report what a succeeded result says.
Capability inventory: The skill includes tools for reading web pages (render_page, scrape, map) and browser automation (unbrowse.browse.*).
Sanitization: No specific sanitization or filtering logic is described within the skill's instructions.
[CREDENTIALS_UNSAFE]: The skill correctly advises users to store the required UNBROWSE_API_KEY in a secret manager rather than committing it to configuration files, which is a security best practice.

python