opper-api

This skill provides direct interface capabilities for the Opper platform via its REST API (api.opper.ai), serving as the foundational layer for all Opper operations. It is designed for developers and engineers building sophisticated LLM applications who need granular control over model routing, performance observation, and infrastructure-level guardrails. By utilizing this skill, you can interact directly with the Opper control plane to manage deployments, execute structured tasks, and implement advanced AI logic without being restricted by high-level abstractions.

The skill covers the complete Opper v3 surface, including the Task API for structured inputs and outputs, as well as the compatibility layer that allows you to swap existing OpenAI, Anthropic, or OpenRouter integrations by simply updating your base URL and API keys. It is the primary tool for addressing architectural queries, understanding API endpoint signatures, and performing model-specific operations such as querying supported providers or pinning specific model versions for consistent behavior.

• Route: Centralize traffic across 200+ models from providers like OpenAI, Anthropic, Google, and Mistral through a single endpoint.

• Observe: Integrate tracing, cost tracking, latency metrics, and evaluation datasets directly into your production pipelines.

• Steer: Enable feedback loops that save high-quality outputs as examples for dataset curation and iterative prompt improvement.

• Guard: Apply infrastructure-level guardrails, including PII removal, content filtering, and budget management before data reaches the model.

• Comply: Meet stringent European data protection directives and security standards, ensuring your LLM architecture is GDPR-ready.

• Always prefer the live OpenAPI specification at https://api.opper.ai/v3/openapi.yaml for definitive documentation on payload schemas and field requirements.

• Use the v3/compat endpoints to replace existing provider-specific SDK logic with Opper's unified gateway, ensuring drop-in compatibility for chat completions and message APIs.

• Access v2 endpoints specifically when dealing with Knowledge Base or RAG-related tasks.

• Authenticate all requests using Bearer token schemes via your Opper platform-issued API key; note that public endpoints like health checks and model listings remain unauthenticated.

• Favor POST /v3/call for native Opper tasks to leverage structured output schemas, native tracing, and multimodal support for image, audio, or video generation.