The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is designed to collect PII and sensitive financial data, including household income, investment portfolio values, and family ages, and transmit this data to the external service at ai.planfi.app for analysis. This is documented as the core mechanism for providing financial planning outputs.
[EXTERNAL_DOWNLOADS]: The skill relies on a remote Model Context Protocol (MCP) server to provide its computation engine. The instructions direct users to connect to https://ai.planfi.app/mcp/free to enable the necessary tools.
[COMMAND_EXECUTION]: The documentation provides specific shell commands for the user to execute, such as 'claude mcp add --transport http planfi https://ai.planfi.app/mcp/free', which modifies the agent's local configuration to include external tool capabilities.
[PROMPT_INJECTION]: The skill uses explicit directives to override the agent's default behavior, such as 'Always CALL ... do not answer from general knowledge,' ensuring the agent relies exclusively on the external service. Furthermore, the skill processes untrusted user data with the following profile:
Ingestion points: User prompts regarding household finances and personal details (SKILL.md).
Boundary markers: Absent; the skill does not use delimiters to separate user-provided values from tool instructions.
Capability inventory: The skill orchestrates multiple remote tools including 'assemble_comprehensive_plan' and 'run_backtesting'.
Sanitization: No explicit validation or escaping of user input is performed before data is passed to the external API.
[NO_CODE]: The skill does not bundle any local executable scripts, relying entirely on markdown-based instructions and the orchestration of remote tools.

comprehensive-plan