WoofAPI OpenAI-Compatible Starter

Minimal Python and Node.js examples for testing an OpenAI-compatible endpoint, plus a small workflow-cost calculator.

Live workflow-cost calculator

Use the free browser tool: https://zcl97630815-cpu.github.io/woofapi-openai-compatible-starter/

It calculates success rate, retry load, cost per successful workflow, monthly forecast, and an optional same-benchmark route comparison. All calculations run locally in the browser.

Migration guide and community

• 30-minute production migration guide: https://zcl97630815-cpu.github.io/woofapi-openai-compatible-starter/openai-compatible-api-migration.html
• Stable release: https://github.com/zcl97630815-cpu/woofapi-openai-compatible-starter/releases/tag/v1.0.0
• Benchmark discussion: #1

The discussion is open for compatibility reports, retry behavior, P95 data, and workflow-cost comparisons. Do not post API keys, private prompts, or customer data.

Why this repository exists

A successful test prompt proves connectivity, not production readiness. Before moving a Dify, RAG, or AI-agent workflow, test:

• streaming and finish reasons;
• JSON output and tool/function calling;
• model-name mappings;
• 401, 404, 429, timeout, and retry handling;
• success rate, P95 latency, and cost per successful workflow.

API endpoint

https://api.woofapi.com/v1

Never commit API keys. All examples read keys from environment variables.

Python quick start

python -m venv .venv
# Windows: .venv\Scripts\activate
# macOS/Linux: source .venv/bin/activate
pip install -r requirements.txt

export WOOFAPI_API_KEY="your-key"
export WOOFAPI_MODEL="your-model-name"
python python/chat.py

PowerShell:

$env:WOOFAPI_API_KEY="your-key"
$env:WOOFAPI_MODEL="your-model-name"
python python/chat.py

Node.js quick start

npm install
export WOOFAPI_API_KEY="your-key"
export WOOFAPI_MODEL="your-model-name"
npm run chat

Calculate workflow cost

python tools/workflow_cost.py --total-cost 18 --successful 82 --attempted 100 --retries 34

Example output:

Success rate: 82.00%
Cost per successful workflow: 0.2195
Retries per attempted workflow: 0.3400

Pricing note

Selected WoofAPI GPT routes can be up to about 95% lower in specific official input-price comparisons. Pricing varies by model, input/output billing, and live route. Check the current pricing page and benchmark your own workload before production.

• Website: https://woofapi.com
• API base URL: https://api.woofapi.com/v1

Disclosure

This starter is maintained for WoofAPI users and examples.

License

MIT