Skip to main content

DEVELOPER DOCUMENTATION

Vibe API key and developer docs.

There is no key issued by Vibe Workflow. Bring a provider key, or use hosted credits. The portal also lists the OpenAPI spec, skills manifest, and MCP server.

// Multi-provider API endpoint
POST https://vibeworkflow.app/api/v1/generate
// Markdown content negotiation
curl -H "Accept: text/markdown" https://vibeworkflow.app/
// MCP server (Streamable HTTP, no auth)
https://vibeworkflow.app/mcp

From intent to implementation

What a Vibe API key is

Vibe Workflow does not issue its own generation key. You bring a key from the AI provider, or you use hosted credits.

The public OpenAPI document sets apiKeyIssuedByProvider to false. A Vibe API key is the provider key you already have for Gemini, OpenAI, Anthropic, OpenRouter, or another supported provider.

On POST /api/v1/generate, send that key in the x-provider-key header or as a bearer token. The app stores the same kind of key in this browser and sends it only for the request that needs it. Hosted AI on Free, Basic, and Pro uses included credits instead. Free includes 50 hosted credits each month. See pricing and how keys are stored.

From intent to implementation

Zero-Friction Agent Onboarding

Instant programmatic access with no sign-up wall. Supply your own API keys for Gemini, OpenAI, Claude, or OpenRouter.

Instant Anonymous Access

No credit card, no account creation, no sales forms. Autonomous agents and developers can probe, query, and generate specifications immediately. The API and MCP server are free — see pricing.

Bring Your Own Key (BYOK)

Pass API keys directly in the x-provider-key header or apiKey request payload. Keys are never stored on server disks.

Standard Rate Limits

Predictable RFC rate limiting with standard response headers (RateLimit-Limit: 60, RateLimit-Remaining, Retry-After).

From intent to implementation

REST & Streaming API Endpoints

Versioned endpoints with typed error schemas (RFC 9457) and multi-provider support.

POST/api/v1/generate
Operation ID: generateTextStream

Streams multi-provider LLM text responses (Gemini, OpenAI, Anthropic, xAI, DeepSeek, Groq, Mistral, OpenRouter) with real-time Google Search Grounding and reasoning effort controls.

POST/api/v1/deep-research
Operation ID: startDeepResearch

Initiates a background multi-turn deep research session using Google Gemini Search Grounding for competitor intelligence and technical feasibility verification.

POST/api/v1/agent-workflow
Operation ID: executeAgentWorkflow

Executes the full 5-stage automated specification synthesis pipeline (Research, PRD, Technical Design, AGENTS.md, Export).

From intent to implementation

Machine-Readable Discovery Resources

Standardized endpoints for AI agents, crawlers, and tool aggregators.

Ready to build your AI specification?

Start planning your next software project with our structured, persona-driven workflow.

Launch Vibe Workflow

Analytics Preferences

Optional analytics help us improve the tool. API keys stay local in your browser. Privacy