The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes 'npx @useclawlink/cli', which triggers the download and execution of the ClawLink CLI package from the NPM registry. This is a standard mechanism for delivering the vendor's functionality.
[COMMAND_EXECUTION]: The skill operates by instructing the agent to execute shell commands using 'npx'. This includes commands for logging in, connecting services, and running specific TinyPNG actions, all of which involve executing external code within the agent's environment.
[INDIRECT_PROMPT_INJECTION]: The skill contains ingestion surfaces where untrusted user data is interpolated into shell commands. Specifically, the commands 'npx @useclawlink/cli actions tinypng ""' and 'npx @useclawlink/cli run tinypng --input ''' (found in SKILL.md) do not specify sanitization or boundary markers. This creates a risk where malicious user input could attempt to manipulate the shell command execution, although the severity is limited by the agent's own safety guardrails.
Ingestion points: User-provided strings for action descriptions and JSON input payloads in SKILL.md.
Boundary markers: None identified.
Capability inventory: Shell command execution via 'npx'.
Sanitization: None explicitly defined in the skill instructions.
[CREDENTIALS_UNSAFE]: The skill documentation notes that authentication results are stored in '~/.clawlink/credentials.json', a sensitive file path that the agent accesses. Furthermore, the 'tinypng_tinify_compress_and_store_in_azure' action requires the use of an Azure SAS token, which is a high-privilege credential that must be handled securely to prevent exposure.

tinypng