The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection because it ingests data from external Render API endpoints and has the capability to perform write operations on the Render platform. * Ingestion points: External data is retrieved via read actions such as render_get_cpu, render_get_memory, and render_get_disk_usage (SKILL.md). * Boundary markers: The instructions include an explicit safety requirement to "confirm with the user before any write" (SKILL.md). * Capability inventory: The skill can execute configuration changes such as render_add_headers, render_add_route, and render_add_or_update_secret_file (SKILL.md). * Sanitization: No specific input sanitization or validation logic is defined within the skill instructions, relying on the underlying CLI and agent behavior.
[COMMAND_EXECUTION]: The skill relies on shell command execution to interact with the Render service through the vendor's command-line interface. * Evidence: The instructions guide the agent to execute subcommands of the @useclawlink/cli tool to perform logins, connections, and actions (SKILL.md).
[EXTERNAL_DOWNLOADS]: The skill downloads and executes the ClawLink CLI from the NPM registry. * Evidence: The skill uses npx @useclawlink/cli commands which fetch the official vendor package from the NPM registry (SKILL.md).

render