Profile My skills Saved Collections Edit profile Submit a skill

← all publishers

kazeizumi

@kazeizumi source repo

6 published skills

  1. ▌
    Skill Governor · kazeizumi bundle
    Govern an agent skill portfolio by discovering installed skills, mapping capabilities and boundaries, reviewing overlaps, recording provenance and task evidence, auditing readiness, and preparing reversible upgrades. Use when skills are missing, duplicated, stale, newly installed, being evaluated, or selected for a multi-stage workflow.
    0
    installs
  2. ▌
    Minimax H3 Cloud · kazeizumi bundle
    Run MiniMax H3 or compatible ComfyUI workflows already installed on an AutoDL application instance. Use to list and select remote workflows, inspect their real input slots, upload references, submit a batch, poll by prompt ID, download results, validate files, and shut the instance down.
    0
    installs
  3. ▌
    H3 Runtime Router · kazeizumi bundle
    Select and control MiniMax H3 execution through local ComfyUI or an AutoDL cloud instance. Use after prompts and references are ready, especially when runtime choice, workflow identity, paid-generation authorization, failure recovery, or cost comparison must be resolved.
    0
    installs
  4. ▌
    Autodl App Instance · kazeizumi bundle
    Securely list, inspect, start, discover, and stop AutoDL Art application instances from Windows, macOS, or Linux. Use when a task needs an AutoDL application instance lifecycle or its proxied ComfyUI panel URL; do not use it to submit generation jobs.
    0
    installs
  5. ▌
    AI Creation Workflow · kazeizumi bundle
    Orchestrate an AI creative project from existing context and project setup through requirement completion, specialist-skill routing, stage dependencies, quality and execution gates, generation, validation, delivery, and reusable evidence. Use for end-to-end or cross-stage creative work rather than one isolated edit.
    0
    installs
  6. ▌
    Comfyui Local Runner · kazeizumi bundle
    Run an existing ComfyUI API-format workflow on a local server from a reproducible JSON manifest. Use after local execution is selected and the exact workflow, input files, overrides, output, and failure fallback are approved.
    0
    installs