The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a custom CLI tool (tinyfish) for web operations. It provides documented commands for search, fetch, agent, and browser modules. These commands are typical for browser automation and web extraction tools.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the CLI via npm install -g @tiny-fish/cli. This is a standard package installation from a public registry. Based on the author context (tinyfish-io), this is a vendor-owned resource and is considered safe within the scope of the skill's primary purpose.
[CREDENTIALS_UNSAFE]: The skill mentions using TINYFISH_API_KEY and an authentication command (tinyfish auth login). It correctly suggests using environment variables rather than hardcoding secrets, which is a safe practice.
[INDIRECT_PROMPT_INJECTION]: The skill processes external web data via fetch and agent commands. While this introduces an ingestion surface for untrusted data, it is a primary function of the skill. The risks are mitigated by the agent's internal safety filters and the clean extraction process that strips boilerplate/scripting from pages by default.
[PRIVILEGE_ESCALATION]: While the skill suggests a global npm install which may require elevated permissions depending on the environment, it does not use sudo or other dangerous privilege escalation techniques within its operational commands.

use-tinyfish