The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the agent to download and execute the ClawLink CLI tool.
Evidence: Commands such as npx @useclawlink/cli login and npx @useclawlink/cli connect retellai trigger external code retrieval via npm.
[COMMAND_EXECUTION]: The skill relies on shell command execution for all its functionality.
Evidence: Multiple CLI commands are provided for authentication, service connection, and Retell AI action execution (e.g., npx @useclawlink/cli run retellai).
[INDIRECT_PROMPT_INJECTION]: The skill processes data from the Retell AI API, which acts as a potential surface for indirect prompt injection.
Ingestion points: The agent reads Retell AI metadata, action descriptions, and execution outputs through the CLI in SKILL.md.
Boundary markers: The instructions lack explicit delimiters or safety instructions to prevent the agent from obeying commands embedded within the Retell AI data.
Capability inventory: The skill utilizes shell command execution and network access via the ClawLink CLI to perform its tasks.
Sanitization: No evidence of data sanitization or integrity checks for the content retrieved from Retell AI is provided.

retellai