The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses npx @useclawlink/cli to download and execute the ClawLink command-line interface directly from the NPM registry.
[COMMAND_EXECUTION]: The skill's primary functionality is delivered through shell commands (login, connect, actions, run), allowing the agent to perform actions in a remote HeyGen account via the ClawLink CLI.
[CREDENTIALS_UNSAFE]: The documentation explicitly identifies ~/.clawlink/credentials.json as the storage location for authentication tokens, which is a sensitive file path.
[INDIRECT_PROMPT_INJECTION]: The skill reads data from external HeyGen actions (e.g., heygen_check_photo_look_generation_status). If the external account data contains instructions, it could influence the agent's behavior.
Ingestion points: Data returned from npx @useclawlink/cli run heygen and describe commands which is then processed by the agent.
Boundary markers: None explicitly defined in the instructions for delimiting or ignoring instructions in the tool output.
Capability inventory: Shell command execution via npx as described in SKILL.md.
Sanitization: None described; the agent is instructed to read the output directly without explicit filtering or validation.
[METADATA_POISONING]: The frontmatter description contains contradictory instructions, suggesting the user paste an API key while the body and setup instructions state no API key is needed due to browser-based OAuth login.

heygen