Krea.ai API

**Quick verdict:** Solid, security-conscious Krea.ai image generation wrapper that fits well into OpenClaw pipelines. It covers the core generate → poll → fetch flow with multiple models and reasonable parameter controls, but it necessarily depends on Krea’s external service quality and policies. **Strengths** - Well-scoped: focuses on Krea image generation with async job handling and polling. - Security-minded: no webhooks, no subprocess shelling, stdlib-only HTTP, and strongly emphasized file permission practices for API keys. - Practical interface: both CLI and Python API (KreaAPI) with clear, concrete usage examples. - Parameter validation helps prevent malformed or abusive requests before hitting the API. **Risks / Limitations** - Hard dependency on Krea.ai’s uptime, latency, rate limits, and model/content policies. - Content safety is largely delegated to Krea’s own safeguards; the skill doesn’t add extra filtering or moderation logic. - No webhook support (by design for security), so long-running jobs rely on polling and may be less efficient for very large batch workflows. - Single-provider design: not a generic multi-backend abstraction; switching away from Krea would require code changes. **Recommended scenarios** - OpenClaw projects that specifically want Krea models (Flux, Imagen 4, Ideogram 3.0, etc.) inside Python or CLI workflows. - Security-sensitive environments that prefer stdlib-only dependencies and explicit credential management. - Prototyping or production pipelines where async generation with polling is acceptable and teams are comfortable with Krea’s service and policies.