The Agent Skills Directory

[SAFE]: The skill implements security-conscious guidance by requiring a strict separation between browser-side search-only keys and server-side write-capable credentials, and by prohibiting the use of admin keys in build variables, bundles, or logs.\n- [SAFE]: Network operations are restricted to fetching official documentation from Algolia, a well-known service, which is considered a safe reference.\n- [INDIRECT_PROMPT_INJECTION]: The skill reads repository content which constitutes untrusted data. (1) Ingestion points: Local repository files accessed via Read, Glob, and Grep (SKILL.md). (2) Boundary markers: Present in the form of 'Tool Discipline' and 'Approval Boundaries' instructions that guide agent behavior. (3) Capability inventory: Write and Edit tools permit modification of local files (SKILL.md). (4) Sanitization: The skill mentions redacted logging but does not define specific sanitization for ingested content. The risk is minimized by the instructional constraints and the requirement for verification before execution.\n- [SAFE]: All identified dependencies are standard, version-pinned packages belonging to the official Algolia ecosystem.

algolia-deploy-integration