The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (LOW): Skill references https://mcp.stripe.com/ for server connectivity. As Stripe is an established and trusted organization, this external reference is considered low risk.
[COMMAND_EXECUTION] (LOW): Provides instructions for adding the MCP server using the claude mcp add command as part of the initial setup.
[PROMPT_INJECTION] (LOW): Susceptible to indirect prompt injection (Category 8) when processing external data. Evidence: 1. Ingestion points: Data entering via search_stripe_documentation and search_stripe_resources tools. 2. Boundary markers: Absent from the prompt instructions. 3. Capability inventory: Includes high-impact write operations such as create_refund and cancel_subscription. 4. Sanitization: Implements a manual confirmation flow for high-risk actions which serves as a human-in-the-loop mitigation.
[NO_CODE] (SAFE): No executable script files (e.g., .py, .js) were included in the skill, reducing the direct attack surface.

mcp-stripe