The Agent Skills Directory

[SAFE]: The skill establishes architectural rules that promote secure coding practices by mandating the use of named client methods (e.g., @agent-native/core/client/hooks). This centralizes cross-cutting concerns like authentication, error handling, and request source validation, reducing the risk of manual implementation errors.
[SAFE]: The included legacy audit (references/legacy-client-fetch-audit-2026-06-03.md) demonstrates a security-conscious design by proactively identifying risks, such as mail integration credentials being stored in browser-readable application state, and providing specific remediation instructions to move them to secure server-side secrets.
[SAFE]: The instructions for the agent (under the 'Don't' section) explicitly forbid generating code that exposes internal routes or duplicates session/auth logic, effectively acting as a defensive guardrail against insecure code generation.

client-methods