feat: Add Forge as LLM provider

[Yiiii0]

Summary

Adds Forge as a new LLM provider in AgentFlow. Uses the forge- model prefix (matching the existing together-, vllm-, ollama- conventions) to route requests through Forge's OpenAI-compatible API.

Changes

• agentflow/.env.template: Added FORGE_API_KEY and FORGE_API_BASE environment variables
• agentflow/agentflow/engine/factory.py: Added forge- prefix detection to route through ChatOpenAI with Forge credentials
• agentflow/agentflow/engine/openai.py: Added optional api_key and base_url parameters to ChatOpenAI.__init__
• assets/doc/api_key.md: Documented Forge API key setup
• assets/doc/llm_engine.md: Added Forge to the LLM engine table

Usage

# Set environment variable
export FORGE_API_KEY="your-forge-api-key"

# Use forge- prefix with Provider/model-name format
python run.py --model "forge-OpenAI/gpt-4o-mini" ...

Test Evidence

All existing tests pass. The change is additive — when not using the forge- prefix, behavior is identical to upstream.

I work at TensorBlock and will help maintain this integration.

---

About Forge

Forge is an open-source middleware service for unified AI model provider management. It routes requests across 40+ AI providers with access to thousands of models through a single OpenAI-compatible API.

• Repo: https://github.com/TensorBlock/forge
• Docs: https://www.tensorblock.co/api-docs/overview