Skip to content

notBlubbll/Stepfun2Opencode

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
AGENTS.md
AGENTS.md
 
 
README.md
README.md
 
 
dashboard.html
dashboard.html
 
 
package.json
package.json
 
 
proxy.js
proxy.js
 
 
proxy.log
proxy.log
 
 
skills.md
skills.md
 
 
start-node.cmd
start-node.cmd
 
 
start.cmd
start.cmd
 
 

Repository files navigation

StepFun2Opencode

OpenAI-compatible proxy server for StepFun AI, providing access to StepFun LLM models through a unified API. Zero external dependencies — uses only Node.js built-in modules.

image image image

Features

  • OpenAI-Compatible API — Standard /v1/chat/completions and /v1/models endpoints
  • Streaming Support — SSE streaming for chat completions
  • Tool Schema Normalization — Resolves $ref and $defs in tool schemas before forwarding
  • Dashboard UI — Liquid glass effects, model/key management, real-time plan & usage stats
  • Auto-Config — Automatically configures opencode provider on startup
  • Dynamic Model Fetch — Fetches available models from https://api.stepfun.ai/step_plan/v1/models
  • Account Info — Real-time balance, cash balance, and voucher balance from StepFun API
  • Step Plan Integration — Live plan name, 5h/weekly usage %, and reset timers from platform.stepfun.ai
  • Response Caching — LRU cache for non-streaming responses
  • Multi-Key Support — Rotate between multiple StepFun API keys
  • Zero Dependencies — No npm packages required

Available Models

Model Description
step-3.5-flash StepFun 3.5 Flash
step-3.5-flash-2603 StepFun 3.5 Flash (2603)
step-3.7-flash StepFun 3.7 Flash (reasoning tiers: low/medium/high)
step-tts-2 StepFun TTS v2
stepaudio-2.5-tts StepFun Audio 2.5 TTS
stepaudio-2.5-asr StepFun Audio 2.5 ASR
step-image-edit-2 StepFun Image Edit v2

Models are dynamically fetched from https://api.stepfun.ai/step_plan/v1/models on startup.

Quick Start

# Clone and start (zero deps — no npm install needed)
cd STEP-PROXY
node proxy.js

# Or use launcher (auto-detects Bun, falls back to Node)
start.cmd

# Open dashboard
open http://localhost:8080

Authentication

Get a StepFun API key from platform.stepfun.ai.

Add to .config/config.json:

{
  "API_KEY": "your-stepfun-api-key"
}

Or set environment variable:

set STEP_API_KEY=your-stepfun-api-key
node proxy.js

Step Plan vs Standard API

StepFun has two separate systems:

Standard API Step Plan
Base URL https://api.stepfun.ai/v1 https://api.stepfun.ai/step_plan/v1
Billing Prepaid balance Subscription quota (5h/weekly limits)
Chat endpoint /v1/chat/completions /step_plan/v1/chat/completions
Models All models step-3.5-flash, step-3.5-flash-2603, step-3.7-flash

The proxy automatically routes chat completions to the Step Plan endpoint. Using the standard endpoint with a Step Plan subscription returns quota_exceeded errors.

Configuration

Edit .config/config.json or set environment variables:

Key Description Default
LISTEN_ADDR Proxy listen address 127.0.0.1:8080
UPSTREAM_BASE_URL StepFun API URL (for models/accounts) https://api.stepfun.ai
API_KEY StepFun API key —
REQUEST_TIMEOUT Upstream request timeout 15m
API_KEYS Client API keys for proxy auth [] (open access)
TOKENS Array of {name, token} for multi-key support auto-populated
OASIS_TOKEN Oasis JWT token for platform.stepfun.ai API —
OASIS_WEBID Oasis web device ID —
CACHE_TTL Response cache TTL 60s
CACHE_MAX_SIZE Max cached responses 100
CACHE_ENABLED Enable response caching true

Getting Session Tokens

The OASIS_TOKEN and OASIS_WEBID are browser session values from platform.stepfun.ai. They are required for the dashboard to display live plan name, 5h/weekly usage %, and reset timers.

  1. Open platform.stepfun.ai in your browser and log in
  2. Open DevTools (F12) → Application tab → Cookies → https://platform.stepfun.ai
  3. Find the Oasis-Token cookie → copy its full value (it's a long JWT string)
  4. Find the Oasis-WebId cookie → copy its full value (a 40-character hex string)

Add to .config/config.json:

{
  "OASIS_TOKEN": "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9...",
  "OASIS_WEBID": "bfe7df2615d72de9fb91c7be9abb5b9105127d5d"
}

Note: The OASIS_TOKEN JWT expires (typically ~30 days). When it expires, the plan status will show an error — regenerate it from the browser. The OASIS_WEBID does not change between sessions.

You can also set per-key sessions in the TOKENS array or via the Dashboard → Manage Keys modal. Per-key sessions override the global OASIS_TOKEN for that key's plan status lookup.

Multi-Key Management

The proxy supports multiple StepFun API keys. Set TOKENS in config:

{
  "TOKENS": [
    { "name": "Key 1", "token": "your-key-1" },
    { "name": "Key 2", "token": "your-key-2" }
  ]
}

Manage keys via the Dashboard → Manage Keys modal (inline add/edit/delete).

Key Rotation & Session Tracking

The proxy automatically rotates tokens across conversations using fingerprint-based session tracking. Each conversation is identified by an MD5 hash of the first user message (skipping auto title prompts). Follow-up requests (tool calls, continuations) in the same conversation are pinned to the same token automatically. A global session counter increments for each new conversation.

Conversation A (fingerprint: a1b2c3) → sess1 → Token 1
├── tool call follow-up → Token 1 (same fingerprint)
└── user follow-up → Token 1 (same fingerprint)
Conversation B (fingerprint: d4e5f6) → sess2 → Token 2 (next in pool)

Console logs show:

14:30:15 [Session#1>Key1]-[step-3.5-flash]-"user prompt"
14:30:15 [Session#1>Key1]-[step-3.5-flash]-done:3996ms

Proxy API Keys

By default the proxy is open access. To restrict access, set API_KEYS:

{
  "API_KEYS": ["my-secret-key-1", "my-secret-key-2"]
}

Clients must include the key:

curl -H "x-api-key: my-secret-key-1" http://localhost:8080/v1/models
curl -H "Authorization: Bearer my-secret-key-1" http://localhost:8080/v1/models

Usage

OpenAI-Compatible

import OpenAI from 'openai';
const client = new OpenAI({
  baseURL: 'http://localhost:8080/v1',
  apiKey: 'not-needed'
});
const response = await client.chat.completions.create({
  model: 'step-3.5-flash',
  messages: [{ role: 'user', content: 'Hello!' }]
});

opencode Integration

The proxy auto-configures opencode on startup. Restart opencode after starting the proxy, then select the stepfun provider.

Dashboard

Access at http://localhost:8080:

  • Plan & Usage Stats — Plan name, 5h/weekly usage % with reset timers (auto-refreshes every 30s)
  • Cache Stats — Real-time cache hits and performance
  • API Key Status — Online/Offline indicator
  • SS Mode — Blur sensitive tokens for screenshots
  • Liquid Glass Effects — Canvas-generated SVG displacement maps
  • Model Management — Toggle models on/off
  • Key Manager — Add/edit/delete API keys with inline editing
  • Collapsible Sections — Models, API Key, Quick Actions, Environment, Proxy Configuration

API Endpoints

Core API

Method Path Description
GET /healthz Health check with API key status, uptime, and account info
GET /v1/models OpenAI models list
POST /v1/chat/completions OpenAI chat completions (streaming supported)

Management API

Method Path Description
GET / POST /api/config Read/write proxy configuration
GET /api/validate Validate API key
GET /api/models List available model IDs
GET /api/bg Bing wallpaper image (cached daily)
GET / POST /api/keys Multi-key CRUD (add/update/delete)
GET /api/account StepFun account info (balance, cash, voucher)
GET /api/step-plan-status Platform plan + usage rates (60s cache)
GET / DELETE /api/cache View/clear response cache

Architecture

proxy.js
├── Config System         — JSON + env vars + API key validation
├── UpstreamClient        — HTTP client for StepFun API
│   ├── getUserInfo()     — GET /v1/models (validate key)
│   ├── chatCompletions() — POST /step_plan/v1/chat/completions
│   ├── getAccountInfo()  — GET /v1/accounts (balance info)
│   ├── getStepPlanStatus()— POST platform QueryStepPlanRateLimit (usage)
│   └── getPlanStatus()   — POST platform GetStepPlanStatus (plan)
├── Tool Schema Norm.     — $ref resolution and schema normalization
├── HTTP Handlers         — OpenAI + management + plan status endpoints
├── Request Router        — Pathname-based routing
├── Session Tracking      — Fingerprint-based sticky sessions
├── Opencode Config       — Auto-configures opencode provider
└── Server Startup        — Validation, config write, listen

dashboard.html
├── Liquid Glass Engine   — Canvas-based displacement/specular maps
├── Plan & Usage Cards    — Plan name, 5h/weekly usage %, reset timers
├── Combined View         — Aggregates plan names and usage from all data slides
├── Model Management      — Toggle models on/off
├── Key Manager           — Add/edit/delete API keys
├── Bing Wallpaper        — Daily rotating background
├── Cache Stats           — Real-time cache performance
└── Configuration Forms   — Listen addr, upstream URL, timeout

Dependencies

No external npm dependencies — uses Node.js built-in modules only: fs, path, os, http, https, url, crypto, zlib.

License

MIT

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

2 stars

Watchers

0 watching

Forks

1 fork
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages