Skills IMA Image Generator

install
source · Clone the upstream repo
git clone https://github.com/openclaw/skills
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/openclaw/skills "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/allenfancy-gan/ima-image-ai" ~/.claude/skills/openclaw-skills-ima-image-generator && rm -rf "$T"
OpenClaw · Install into ~/.openclaw/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/openclaw/skills "$T" && mkdir -p ~/.openclaw/skills && cp -r "$T/skills/allenfancy-gan/ima-image-ai" ~/.openclaw/skills/openclaw-skills-ima-image-generator && rm -rf "$T"
manifest: skills/allenfancy-gan/ima-image-ai/SKILL.md
source content

IMA Image AI

When To Use

Use this repository when the user wants an image output:

  • text-to-image
  • image-to-image
  • style transfer
  • continuity via reference image

This repo is image-only. Do not route video generation, audio generation, or non-image tasks here.

Gateway Contract

  • New-machine bootstrap entrypoint:
    python3 scripts/ima_runtime_setup.py
  • Environment/self-check entrypoint:
    python3 scripts/ima_runtime_doctor.py
  • The official CLI entrypoint is
    python3 scripts/ima_runtime_cli.py ...
    .
  • Always query the live product list before task creation so
    attribute_id
    ,
    credit
    ,
    model_version
    , and defaults come from the current catalog.
  • Return remote HTTPS image URLs; do not download results into local attachments for the user.
  • Route requests through the image capability so
    text_to_image
    and
    image_to_image
    are classified before execution.
  • If
    --model-id
    is omitted, the runtime uses the recommended default model for that task type.
  • Auto-selected defaults are operational fallbacks, not persisted user preferences.

Quick Start

  • Minimal path:
    1. python3 scripts/ima_runtime_setup.py --install
    2. export IMA_API_KEY="ima_your_key_here"
    3. python3 scripts/ima_runtime_cli.py --task-type text_to_image --prompt "a cinematic mountain sunset" --output-json
  • Use
    python3 scripts/ima_runtime_doctor.py --output-json
    when setup passes but runtime or catalog access still fails.

Operator References

  • README.md
    covers first-use paths and canonical entry commands.
  • references/shared/catalog-aware-selection.md
    defines the formal live-catalog-aware model-selection contract.
  • references/operations/troubleshooting.md
    covers common failure recovery.
  • capabilities/image/references/parameter-tuning.md
    covers
    size
    ,
    aspect_ratio
    , and
    n
    usage.
  • capabilities/image/references/scenarios.md
    covers prompt-only and reference-image examples.

Read Order

  1. references/README.md
  2. references/gateway/entry-and-routing.md
  3. references/gateway/workflow-confirmation.md
  4. references/shared/model-selection-policy.md
  5. references/shared/catalog-aware-selection.md
  6. references/shared/error-policy.md
  7. references/shared/security-and-network.md
  8. references/operations/troubleshooting.md
  9. capabilities/image/CAPABILITY.md
  10. capabilities/image/references/parameter-tuning.md
  11. capabilities/image/references/scenarios.md

Boundary

  • references/gateway/*
    covers entry, routing, and clarification seams.
  • references/shared/*
    covers rules reused across the runtime.
  • capabilities/image/*
    owns image-specific behavior.
  • _meta.json
    and
    clawhub.json
    are metadata inputs, not the primary narrative docs.