No user-submitted related servers found.
Scored across 11 tools
Most tools have a genuinely distinct role: single video, image-to-video, still image, and batch queueing are cleanly separated by input type and scope. The main soft spots are flow_check_session vs flow_health_check (both readiness probes, differing only in depth) and flow_preview vs flow_get_queue_status, though the description explicitly redirects the model to the latter. These are documented overlaps rather than true collisions.
All names share a uniform flow_ prefix with lowercase snake_case, which reads predictably. Minor deviations: flow_health_check uses noun_verb ordering while flow_check_session uses verb_noun, and flow_preview/flow_cancel lack an explicit object. Still easy to scan and group.
Eleven tools is well within a healthy range for a browser-automation generation server. Each generation variant earns its place (text-to-video, frames/ingredients, stills, batch), and the supporting set (cost estimate, session/health check, queue status, cancel, file open) maps to real operational needs.
The lifecycle is covered end-to-end: verify session, estimate cost, generate (single and batch), poll status, cancel, and open results. Gaps are minor — no tool to read back a manifest/batch results on disk, no listing of prior jobs or projects, and no cleanup/delete for finished work, so agents must fall back to the filesystem for those.