The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the prism CLI for organization management and signing-key operations. Specific commands include prism me, prism signing-key generate, and prism signing-key import. The instructions explicitly mandate using the --private-key-file flag to ensure private keys are saved to disk rather than being printed to logs, console output, or the agent's response.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the @prismatic-io/prism and @prismatic-io/embedded packages from the npm registry. These are official tools and SDKs provided by the skill's author to facilitate the embedding process.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes external host application code and user requirements to generate tailored integration logic, which represents an indirect injection surface.
Ingestion points: Host application source code (frontend and backend), configuration files, and user-provided requirements.
Boundary markers: The agent is directed to prioritize existing application patterns (auth middleware, routers) and infer technical choices directly from the code while prompting the user for business-level decisions.
Capability inventory: The skill generates implementation code and executes prism CLI commands for platform configuration.
Sanitization: The instructions emphasize security by requiring that sensitive identity claims (sub, customer) be sourced only from authenticated server-side sessions, effectively ignoring untrusted client-side inputs.

prismatic-embedded