From 63dd26aae29f02e42fcb4ebb873f146122102ed6 Mon Sep 17 00:00:00 2001 From: hei Date: Wed, 15 Jul 2026 01:28:39 +0800 Subject: [PATCH] docs: add DaoXE OpenAI-compatible gateway example Document using DaoXE (https://daoxe.com/v1) via LLM_PROVIDER=openai and LLM_API_BASE. Model IDs are account-scoped; note multi-protocol scope and mainland China availability. Co-Authored-By: Claude Fable 5 --- legacy/.env.template | 3 ++- legacy/README.md | 13 +++++++++++++ 2 files changed, 15 insertions(+), 1 deletion(-) diff --git a/legacy/.env.template b/legacy/.env.template index 2552f4538..dea8020a7 100644 --- a/legacy/.env.template +++ b/legacy/.env.template @@ -30,7 +30,8 @@ LLM_TEMPERATURE=0.2 LLM_MAX_TOKENS=4096 LLM_TIMEOUT=60 -# Optional OpenAI-compatible overrides +# Optional OpenAI-compatible overrides (e.g. DaoXE multi-model gateway: https://daoxe.com/v1) +# Model IDs are account-scoped; service not available in mainland China. LLM_API_BASE= LLM_API_VERSION= diff --git a/legacy/README.md b/legacy/README.md index 102c5b8fc..36ab33f9f 100644 --- a/legacy/README.md +++ b/legacy/README.md @@ -88,6 +88,19 @@ Core libraries live at the repo root under `potpie/` (`context-engine`, `parsing > **`CHAT_MODEL`** and **`INFERENCE_MODEL`** are used for agent reasoning and knowledge graph generation respectively. Model names follow the `provider/model_name` format as expected by [LiteLLM](https://docs.litellm.ai/docs/providers). > > **💡 Using Ollama instead?** Set `LLM_PROVIDER=ollama` and use `CHAT_MODEL=ollama_chat/qwen2.5-coder:7b` and `INFERENCE_MODEL=ollama_chat/qwen2.5-coder:7b`. + > + > **Using an OpenAI-compatible gateway (e.g. [DaoXE](https://daoxe.com))?** DaoXE is a multi-model multi-protocol API gateway (`https://daoxe.com/v1`). Point the OpenAI path at it with your DaoXE key and an account-scoped model ID (from authenticated `GET /v1/models` or your dashboard — not a hardcoded public catalog). DaoXE is not available in mainland China. + > + > ```bash + > LLM_PROVIDER=openai + > OPENAI_API_KEY=your-daoxe-api-key + > # Account-scoped model IDs, e.g. openai/ + > CHAT_MODEL=openai/your-account-model-id + > INFERENCE_MODEL=openai/your-account-model-id + > LLM_API_BASE=https://daoxe.com/v1 + > ``` + > + > DaoXE also exposes other protocol routes (for example Anthropic Messages) for clients that speak those APIs; with Potpie's OpenAI/`LLM_API_BASE` path above you use Chat Completions. Starters: [DaoXE-AI](https://github.com/seven7763/DaoXE-AI) · [Pricing](https://daoxe.com/pricing). See `.env.template` for the full list of optional configuration (logging, feature flags, object storage, email, analytics, etc.).