The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a local CLI smart-search for all web research and configuration tasks, requiring the binary to be present in the user's PATH. It also features a 'Deep Research' mode where the agent generates and executes sequences of shell commands based on user queries.\n- [CREDENTIALS_UNSAFE]: The skill provides commands (smart-search config set) to manage sensitive API keys for various search providers. These are stored in local JSON configuration files. While the doctor command masks secrets, other listing commands like config list may expose them in raw output.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection.\n
Ingestion points: Web content is ingested via fetch and search subcommands in SKILL.md.\n
Boundary markers: Output is formatted as JSON or Markdown, but there are no explicit instructions to ignore embedded prompts in the fetched text.\n
Capability inventory: The skill can execute various subcommands of the smart-search local binary (search, fetch, exa-search, zhipu-search, map), including multi-step 'Deep Research' orchestration.\n
Sanitization: No sanitization of external web content is mentioned before it is processed by the agent.

smart-search-cli