The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses npx @useclawlink/cli to perform actions such as logging in, connecting services, and running specific Linkhut actions.
[EXTERNAL_DOWNLOADS]: The skill downloads and executes the @useclawlink/cli package from the NPM registry using the npx utility.
[CREDENTIALS_UNSAFE]: The skill documentation notes that user credentials for the service are stored at ~/.clawlink/credentials.json. This is standard behavior for the vendor's CLI tool and is documented for transparency.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from Linkhut, which is an external source and could potentially contain untrusted instructions.
Ingestion points: Untrusted data enters the agent context via the output of Linkhut actions (e.g., reading bookmark data).
Boundary markers: The skill includes explicit instructions for the agent to "Reads first; confirm with the user before any write," which serves as a manual boundary.
Capability inventory: The skill has the capability to execute shell commands and interact with external APIs through the vendor CLI.
Sanitization: No automated sanitization of external Linkhut content is described in the prompt instructions.

linkhut