The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill recommends the installation of the @t2000/cli package from the NPM registry, which is a third-party dependency not belonging to a known trusted organization.\n- [REMOTE_CODE_EXECUTION]: The skill configures the agent to use a remote MCP server at https://mcp.t2000.ai/mcp. This allows the agent's tools and behaviors to be defined and updated by an external service provider.\n- [COMMAND_EXECUTION]: The instructions include shell commands like npm i -g @t2000/cli and npx @t2000/cli mcp uninstall for software installation and maintenance, which involves executing code on the user's system.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to retrieve and process user-generated job descriptions from an external job board, creating a surface for indirect instructions.\n
Ingestion points: The t2000_job_status tool fetches the workOrder field, which contains untrusted content from the t2000 service.\n
Boundary markers: There are no instructions or delimiters provided to the agent to distinguish between data and potential commands within the workOrder.\n
Capability inventory: The agent has access to financial tools such as t2000_job_settle, pay x402, and send. A malicious workOrder could attempt to manipulate the agent into misusing these capabilities.\n
Sanitization: There is no mention of content filtering or validation for the job data retrieved from the server.

t2000-connect