Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
151 changes: 151 additions & 0 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -363,6 +363,157 @@ VoltAgent is a community effort, and it keeps getting better because of people l

![Contributors](https://contrib.rocks/image?repo=voltagent/voltagent&max=500&columns=20&anon=1)

## ❓ FAQ

### What is VoltAgent?

VoltAgent is an end-to-end **AI Agent Engineering Platform** that provides:
- **Open-Source TypeScript Framework**: Build agents with memory, tools, workflows, voice, and RAG
- **VoltOps Console**: Production-ready observability, automation, deployment, and evaluations

### How is VoltAgent different from other agent frameworks?

| Feature | VoltAgent | LangChain | CrewAI |
|---------|-----------|-----------|--------|
| **Architecture** | TypeScript-first | Python-first | Python-first |
| **Workflow Engine** | Declarative workflows | Sequential chains | Role-playing teams |
| **Memory** | Multiple adapters | In-memory + options | Shared memory |
| **Observability** | Built-in VoltOps Console | External tools | Limited |
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2: The comparison table contains inaccurate characterizations of competitor capabilities:

  • LangChain's workflow engine is not just "Sequential chains" — LangGraph provides stateful, graph-based workflows with agent patterns.
  • LangChain's observability is not "External tools" — LangSmith is their official built-in observability platform with tracing, monitoring, and dashboards.
  • CrewAI's observability is not "Limited" — CrewAI AMP provides built-in tracing with OpenTelemetry and performance metrics.
  • CrewAI's workflow engine is not just "Role-playing teams" — Flows support @start, @listen, @router for multi-step pipelines.

Publishing misleading comparisons undermines credibility. Consider updating to reflect current capabilities accurately.

Prompt for AI agents
Check if this issue is valid — if so, understand the root cause and fix it. At README.md, line 381:

<comment>The comparison table contains inaccurate characterizations of competitor capabilities:
- LangChain's workflow engine is not just "Sequential chains" — LangGraph provides stateful, graph-based workflows with agent patterns.
- LangChain's observability is not "External tools" — LangSmith is their official built-in observability platform with tracing, monitoring, and dashboards.
- CrewAI's observability is not "Limited" — CrewAI AMP provides built-in tracing with OpenTelemetry and performance metrics.
- CrewAI's workflow engine is not just "Role-playing teams" — Flows support `@start`, `@listen`, `@router` for multi-step pipelines.

Publishing misleading comparisons undermines credibility. Consider updating to reflect current capabilities accurately.</comment>

<file context>
@@ -363,6 +363,157 @@ VoltAgent is a community effort, and it keeps getting better because of people l
+| **Architecture** | TypeScript-first | Python-first | Python-first |
+| **Workflow Engine** | Declarative workflows | Sequential chains | Role-playing teams |
+| **Memory** | Multiple adapters | In-memory + options | Shared memory |
+| **Observability** | Built-in VoltOps Console | External tools | Limited |
+| **Deployment** | One-click cloud | Manual setup | Manual setup |
+| **Voice Support** | Built-in TTS/STT | External providers | No |
</file context>

| **Deployment** | One-click cloud | Manual setup | Manual setup |
| **Voice Support** | Built-in TTS/STT | External providers | No |

Comment on lines +376 to +384
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Potential issue | 🟠 Major | ⚡ Quick win

🧩 Analysis chain

🌐 Web query:

Verify current public docs for LangChain and CrewAI feature support (workflow model, observability, voice) and confirm listed model examples for OpenAI/Anthropic/Google/Groq are still accurate as of May 2026.

💡 Result:

LangChain and CrewAI both support workflow models, observability, and voice features as of May 2026, per their official docs. Listed model examples for OpenAI (e.g., gpt-5.x), Anthropic (e.g., claude-sonnet-4-6), Google (e.g., gemini-2.5/3.x), and Groq (via langchain-groq integration with models like llama-3.3-70b) remain accurate and current.[1][2][3][4][5] LangChain: - Workflow model: LangGraph enables stateful, graph-based workflows with nodes, edges, persistence, and agent patterns.[6][7][8] - Observability: LangSmith provides tracing, monitoring, dashboards, alerts, feedback, and integrations.[9][10] - Voice: Official docs detail voice agents using STT + LangChain agent + TTS (e.g., AssemblyAI, Cartesia).[11] CrewAI: - Workflow model: Flows with @start, @listen, @router for multi-step pipelines, state management, crews integration.[12][13][14][15] - Observability: Built-in tracing (CrewAI AMP), OpenTelemetry, performance metrics, LangSmith/CrewAI integrations.[16][17] - Voice: No explicit native voice support found in docs; general LLM streaming possible but not voice-specific.[18][19] Model support confirmed via LangChain docs (init_chat_model with provider prefixes) and Groq integration (ChatGroq).[1][2][3][5] CrewAI inherits via LLM flexibility (e.g., OpenAI/gpt-4o-mini).[13]

Citations:


Update comparison table with current capabilities; several LangChain and CrewAI claims are now inaccurate.

The table at lines 376-384 understates LangChain and CrewAI capabilities:

  • LangChain "Sequential chains": LangGraph now provides stateful, graph-based workflows with agent patterns, not just sequential execution.
  • LangChain "External tools" (Observability): LangSmith is the official observability platform with tracing, monitoring, dashboards, and alerts—not merely external options.
  • CrewAI "Limited" (Observability): CrewAI AMP provides built-in tracing with OpenTelemetry and performance metrics; this is understated.
  • CrewAI "Role-playing teams" (Workflow Engine): Flows support @start, @listen, @router for multi-step pipelines beyond role-based teams.

CrewAI's "No" for voice support remains accurate. Model examples (gpt-5.x, claude-sonnet-4-6, gemini-2.5/3.x, llama-3.3-70b) are current as of May 2026.

Also applies to: 395-400

🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@README.md` around lines 376 - 384, The comparison table currently misstates
LangChain and CrewAI capabilities: update the LangChain row (the "Sequential
chains" and "External tools" cells) to reflect stateful, graph-based workflows
and agent patterns (e.g., LangGraph) and to list LangSmith as built-in
observability with tracing/monitoring/dashboards/alerts; update the CrewAI row
(the "Role-playing teams" and "Limited" observability cells) to reflect Flows'
`@start/`@listen/@router multi-step pipelines and built-in AMP observability with
OpenTelemetry tracing and performance metrics while keeping CrewAI's "No" for
Voice Support; also ensure the second table instance noted (lines ~395-400)
receives the same corrections and keep the model examples line current (gpt-5.x,
claude-sonnet-4-6, gemini-2.5/3.x, llama-3.3-70b).

### How do I install VoltAgent?

```bash
npm create voltagent-app@latest
```

This creates a new project with starter code and examples.

### Which LLM providers are supported?

VoltAgent supports all major providers through AI SDK:
- **OpenAI**: GPT-4, GPT-4o, o1, o3-mini
- **Anthropic**: Claude 3.5, Claude Sonnet 4
- **Google**: Gemini 1.5 Pro, Gemini 2.0 Flash
- **Groq**: Llama, Mixtral (fast inference)
- **Custom**: Any provider supported by AI SDK

### What are Supervisors and Sub-Agents?

VoltAgent supports **multi-agent systems** where:
- **Supervisor**: Routes tasks to specialized sub-agents
- **Sub-Agents**: Handle specific tasks (e.g., weather, expenses, research)

Example: A supervisor agent coordinates weather lookup and expense approval agents.

### How do I add custom tools?

Use the `@voltagent/core` tool registry:

```typescript
import { VoltAgent, Agent } from "@voltagent/core";
import { weatherTool } from "./tools";

const agent = new Agent({
name: "Weather Agent",
tools: [weatherTool],
});
```

### What is MCP support?

VoltAgent supports **Model Context Protocol (MCP)** servers:
- Connect to MCP servers without extra glue code
- Use existing MCP tools and resources
- Example: File system MCP, database MCP, custom MCP

### How does memory work?

VoltAgent provides **durable memory adapters**:
- **LibSQL**: SQLite-based persistent storage
- **In-Memory**: Default, no persistence
- Custom adapters possible

### What is VoltOps Console?

VoltOps Console is the **production operations platform**:
- **Observability**: Trace agent runs, view logs
- **Automation**: Schedule workflows, set triggers
- **Deployment**: One-click cloud deployment
- **Evals**: Run evaluation suites
- **Guardrails**: Enforce content policies

Options: Cloud (managed) or Self-Hosted.

### How do I use RAG?

VoltAgent provides built-in RAG capabilities:

```typescript
import { VoltAgent, Agent } from "@voltagent/core";
import { retrieverAgent } from "./retriever";

const agent = new Agent({
name: "RAG Agent",
retriever: retrieverAgent,
});
```

Or use **VoltAgent Knowledge Base** for managed RAG service.

### How do I add voice support?

VoltAgent provides built-in voice providers:

```typescript
import { VoltAgent, Agent } from "@voltagent/core";
import { openai } from "@voltagent/voice-openai";

const agent = new Agent({
name: "Voice Agent",
voice: { provider: openai() },
});
```

### How do I run evaluations?

Use the Evals system:

```typescript
import { EvalSuite } from "@voltagent/evals";

const suite = new EvalSuite({
name: "Weather Agent Eval",
tests: [...],
});

await suite.run();
```

### What guardrails are available?

VoltAgent provides runtime guardrails:
- **Input Guardrails**: Validate agent input
- **Output Guardrails**: Validate agent output
- Content policy enforcement
- Custom validation rules

### Can I use VoltAgent without VoltOps Console?

Yes! The **core framework** is open-source and works standalone:
- Build agents with all features
- Deploy manually or use VoltOps Console later
- Console is optional for production operations

### Where can I get help?

- 📖 [Documentation](https://voltagent.dev/docs/)
- 💬 [Discord Community](https://s.voltagent.dev/discord)
- 🐛 [GitHub Issues](https://github.com/voltagent/voltagent/issues)
- 📧 [Twitter](https://x.com/voltagent_dev)


## License

Licensed under the MIT License, Copyright © 2026-present VoltAgent.