The Agent Skills Directory

[SAFE]: The skill instructions promote secure coding practices by requiring explicit user credentials and emphasizing the use of official vendor libraries.
[COMMAND_EXECUTION]: Contains a discovery command using curl in SKILL.md to fetch index metadata from Algolia's API. The command is benign, uses placeholders for sensitive information, and targets a trusted vendor domain (algolia.net).
[EXTERNAL_DOWNLOADS]: Guides the agent to fetch live documentation and install verified packages from official registries, adhering to the trust scope rules for well-known services.
[INDIRECT_PROMPT_INJECTION]: The skill describes a process for ingesting external schema data in SKILL.md to discover attributes. This risk is mitigated by explicit instructions to confirm the discovered schema with the user before proceeding, maintaining a human-in-the-loop.

instantsearch