ai-provider-openai-compat

Enterprise Part of the commercial Kumiko Enterprise offering — not in the open framework.

Registers an OpenAI Chat-Completions-compatible backend as the "openai-compat" LLM provider in ai-foundation. A single adapter covers OpenAI, Ollama, vLLM, llama.cpp, OpenRouter, and other Chat-Completions-speaking endpoints — point the per-tenant endpoint config-key at the server you want and set the model name; the API key is optional for local backends. No OpenAI SDK dependency; direct fetch.

How it fits

What this feature needs to run (Requires, top) and the write commands it provides (Provides, bottom).

flowchart TB
  n_ai_provider_openai_compat["ai-provider-openai-compat"]
  subgraph how_reqs["Requires"]
    n_config["config"]
    n_secrets["secrets"]
    n_ai_foundation["ai-foundation"]
  end
  n_config --> n_ai_provider_openai_compat
  n_secrets --> n_ai_provider_openai_compat
  n_ai_foundation --> n_ai_provider_openai_compat

Dependencies

• Requires: config, secrets, ai-foundation
• Activation: always on (not toggleable)

Configuration

Per-tenant config keys, set via the tenant-admin UI or a seed. 🔒 = encrypted at rest.

Key	Type	Default	Scope	Who can write	Who can read
endpoint	text	""	tenant	TenantAdmin, SystemAdmin	TenantAdmin, SystemAdmin
max-tokens	number (1024–128000)	16000	tenant	TenantAdmin, SystemAdmin	all
model	text	""	tenant	TenantAdmin, SystemAdmin	all

Secrets

Secret	Scope	Hint
ai-provider-openai-compat:secret:openai-compat-api-key	tenant	Optional. Required for OpenAI/OpenRouter, leave empty for local Ollama/vLLM.

Extensions & cross-feature APIs

• Registers extension: llmProvider → openai-compat