The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references and fetches information from Plaid's official documentation and API endpoints (e.g., plaid.com, sandbox.plaid.com) to facilitate testing.
[INDIRECT_PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection by accepting untrusted user inputs that are subsequently passed to network-enabled tools.
Ingestion points: The skill ingests user-provided client_id, secret, and webhook URLs as required inputs within the SKILL.md workflow.
Boundary markers: There are no explicit delimiters or specific warnings instructed to isolate these user-provided strings within the agent's prompts.
Capability inventory: The skill leverages Model Context Protocol (MCP) tools such as plaid_createSandboxItem, plaid_getAccounts, and plaid_fireSandboxWebhook which perform network requests based on the ingested data.
Sanitization: The instructions do not specify any validation, filtering, or escaping mechanisms for the external inputs before they are used in tool calls.

plaid-sandbox-testing