The Agent Skills Directory

[SAFE]: The skill implements robust security controls for interacting with the external Bria API and avoids common pitfalls in network-enabled tools.- [EXTERNAL_DOWNLOADS]: The skill downloads generated images from remote hosts. It mitigates SSRF and malicious file risks by enforcing HTTPS, blocking redirects, requiring a user-provided host allowlist, and validating image magic numbers and pixel counts using the Pillow library before moving files to a permanent location.- [SAFE]: Cost control and authorization are strictly enforced via environment variables and request hashing. The provided client script requires explicit human approval of the request hash and a pre-calculated cost ceiling before making any quota-consuming network requests.- [SAFE]: Authentication tokens (api_token) are handled securely via environment variables and are explicitly prevented from being logged, committed to source, or forwarded to untrusted status or asset hosts.- [SAFE]: Job state management and artifact storage use atomic file operations, such as O_EXCL for lock files and os.replace for temporary files, to ensure data integrity and prevent race conditions.

bria-fibo-image