OpenClaw alternative · Compared honestly · Public Alpha

An OpenClaw alternative with memory in your browser, your editor and your chats.

Stop re-explaining your work.

Choose OpenClaw if you want a free, MIT-licensed assistant with no account, 32 listed messaging channels, phone apps and a community of 389,000+ GitHub stargazers. Choose Vodou, the OpenClaw alternative on this page, if the memory has to reach past your chats: Vodou types it into ChatGPT, Claude and 20 other AI sites in Chrome, loads it into Claude Code and Cursor through hooks, and shares it with editors through an MCP host, while skills run their steps in a fixed order. Vodou needs a free account, costs $10 a month on your own model keys after a 60-day trial, and its engine is proprietary. Vodou imports an OpenClaw workspace's memory with one command.

Download Vodou See the comparison

Imports OpenClaw memory22 AI sites in ChromeClaude Code · Cursor hooks60-day free trial

TL;DR

Vodou and OpenClaw both run on your own computer; OpenClaw is free, MIT-licensed, needs no account and lists 32 messaging channels, while Vodou costs $10 a month on your own keys after a 60-day trial and requires a free account.
Vodou types its memory into 22 AI sites in Chrome, including ChatGPT and Claude, and loads it into Claude Code through 4 hooks and into Cursor; OpenClaw has no chat-site extension and no Claude Code or Cursor hooks.
Vodou's MCP host gives Claude Desktop, Cursor, VS Code, Windsurf and Zed their own token, memory vault and rate limit on 127.0.0.1:8787; OpenClaw's openclaw mcp serve shares its channel conversations with MCP clients.
Vodou imports an OpenClaw workspace — MEMORY.md, USER.md and every note under memory/ — with vodou-core mem import openclaw, previewed first with --dry-run and reversible with mem import-undo.

Definition

What is OpenClaw?

OpenClaw is a free, open-source personal AI assistant that runs on your own computer and answers you in the chat apps you already use. One local Gateway connects hosted or local models to WhatsApp, Telegram, Slack, Discord, iMessage and 27 more listed channels, with ClawHub skills, a Markdown memory and native apps for macOS, iOS, Android, Windows and Linux.

OpenClaw is stewarded by the OpenClaw Foundation, an independent 501(c)(3), and its README says the project has “no paid tier, hosted service, or token.” It is MIT-licensed, installs with one script on macOS, Linux and Windows, and had 389,000+ GitHub stars on September 17, 2026, with release v2026.9.4 out on September 11. The same Gateway can run as one person's assistant on a laptop or as a shared team deployment.

OpenClaw keeps memory as Markdown files in its workspace — MEMORY.md for durable facts, USER.md for preferences and dated notes under memory/ — indexed in SQLite with keyword, vector and hybrid search; its docs say it uses OpenAI embeddings by default, with other embedding providers and Honcho or LanceDB engines as options. Vodou also keeps memory as Markdown files indexed in SQLite, which is why Vodou can import an OpenClaw workspace file for file.

The problem

Why do people look for an OpenClaw alternative?

The reasons in 2026's alternative guides and Reddit threads are consistent: security after the ClawHavoc skill-registry attack and CVE-2026-25253, memory that forgets or mixes projects, and the upkeep of a self-hosted gateway that updates often. Kilo's analysis of about 1,300 Reddit comments names memory failures as the top reason people leave. Vodou is one answer to the memory reason; the security and upkeep reasons apply to any agent you run yourself, Vodou included.

Security after ClawHavoc

Researchers found 341 malicious skills on ClawHub in early 2026, and CVE-2026-25253 (CVSS 8.8) allowed one-click remote code execution until v2026.1.29 fixed it. OpenClaw runs tools on the host unless you turn on its sandbox, which is off by default.

Memory that forgets

Reddit users describe agents forgetting instructions and mixing data between projects. OpenClaw's memory is Markdown per workspace plus a SQLite index, so what matters is whether the right note is found and loaded at the right moment.

Upkeep and token bills

Guides cite the Node.js setup, frequent releases that can break a working gateway, and token costs that grow with long histories. Moving to another agent you run yourself, Vodou included, moves that upkeep rather than removing it.

Capabilities

How is Vodou different from OpenClaw?

Vodou is built around the memory rather than the chat channel. OpenClaw centres on a Gateway that puts one assistant in 32 listed messaging apps; Vodou centres on a local memory that the Vodou Bridge extension, Claude Code and Cursor hooks, an MCP host and your messaging channels all read and write, plus skills whose steps run in order. Memory is stored and searched on your computer; with a hosted model, what you send goes to that provider.

Memory in ChatGPT, Claude and 20 more AI sites

The Vodou Bridge extension for Chrome types relevant memory into the chat box on 22 AI sites when you press Ctrl+B, where you can see and edit it before you send — inject is manual by default, so nothing is sent on your behalf. It also captures those chats back into memory.

Memory in Claude Code and Cursor

Four hooks give Claude Code memory at session start and on every prompt, and save what it learned when the session ends. In Cursor, memory loads at session start and every prompt refreshes a context file of matching memories that the Cursor agent reads.

An MCP host for every editor

Claude Desktop, Cursor, VS Code, Windsurf and Zed attach to Vodou on 127.0.0.1:8787, each with its own token, memory vault and rate limit, and a 30-day audit log records what each client read.

Skills that run their steps in order

A Vodou skill is a SKILL.md file whose AGENT_ACTIONS block the workflow driver executes step by step, stopping where the skill says to ask you, so the model cannot skip or reorder steps. Add as many skills as you like; a Skill Console gives one its own chat tab and schedule.

Any model, never metered

Bring keys for any model provider, or run Ollama, LM Studio or the bundled llama.cpp, and switch per conversation. With a local model nothing you say leaves your machine; with a hosted one, what you send — including the memory extraction step — goes to that provider.

Memory you can correct and undo

Every memory in Vodou keeps its source. Correct a wrong fact once, pin what matters, hide everything learned from one website, or undo a whole import — and every surface that reads the memory gets the fix.

Comparison

Vodou vs OpenClaw: how do they compare?

OpenClaw wins on price, license, account-free setup, community size, built-in channel count, phone apps and sandbox options; Hermes Agent, the alternative most guides list first, shares most of those strengths. Vodou's column is about reach: the same local memory in 22 browser AI sites, Claude Code, Cursor and an MCP host with a vault per client, skills that run in a fixed order, and an importer for OpenClaw memory. Vodou is a public alpha.

→Scroll the table sideways to see 2 columnsall 3 columns

CAPABILITY	VODOU public alpha	OPENCLAW	HERMES AGENT the alternative most guides list first
What it is	local memory layer + agent harness	self-hosted personal assistant (OpenClaw Foundation)	self-improving agent (Nous Research)
Price	$10/mo on your own keys after a 60-day free trial	free — “no paid tier, hosted service, or token”	free; optional Nous Portal subscription for models and hosted tools
Account required	yes — a free Vodou account licenses the engine	no	no (Nous Portal sign-in only if you use it)
License	open-core: gateway, skills, extension, MCP servers Apache-2.0; engine proprietary	MIT	MIT
GitHub stars (Sep 17, 2026)	public alpha — early community	389,000+	246,000+
Messaging channels	Telegram, Slack, Discord, WhatsApp, Signal, iMessage, Teams, Google Chat, voice, web — plus any channel package you install	32 listed, incl. WhatsApp, Telegram, Slack, Discord, iMessage, Signal, Teams, Matrix, LINE, WeChat	29 listed, incl. Telegram, Discord, Slack, WhatsApp, Signal, Email, SMS, Teams, Matrix, iMessage via BlueBubbles
Apps	web console and CLI; no phone apps — reach it from your phone through a channel	native apps for macOS, iOS, Android, Windows, Linux + Control UI	Hermes Desktop (macOS, Windows), terminal UI, CLI
Models	any provider on your keys; Ollama, LM Studio, bundled llama.cpp	hosted and local providers; Claude and Codex as harness plugins	Nous Portal (300+ models), OpenRouter, OpenAI, your own endpoint
Memory store	Markdown logs + SQLite: full-text, on-device vectors, reranker; provenance; correct, pin, undo	MEMORY.md, USER.md, daily notes; SQLite keyword + vector hybrid; OpenAI embeddings by default; Honcho or LanceDB engines	MEMORY.md (2,200 chars) + USER.md (1,375 chars); FTS5 session search; 8 memory provider plugins
Memory in browser AI chats	22 sites via the Vodou Bridge Chrome extension	no chat-site extension	no chat-site extension
Memory in Claude Code / Cursor	hooks: Claude Code (4 hooks); Cursor (session start + per-prompt context file)	no hooks; runs Claude and Codex as harness plugins instead	no hooks; hermes mcp serve lends Claude Code, Cursor and Codex its messaging tools
MCP	client for any server; host on 127.0.0.1:8787 with per-client token, vault, rate limit, audit	client (openclaw mcp add); openclaw mcp serve shares channel conversations	client (stdio, HTTP, OAuth); hermes mcp serve exposes 10 messaging tools
Skills	SKILL.md + AGENT_ACTIONS run in order; imports Hermes-format skills	ClawHub community registry + plugins	writes and improves its own; agentskills.io compatible
Scheduled and background work	60-second scheduler, automations, kanban board (worker processes opt-in)	cron jobs, background tasks, sub-agents	cron with delivery to any platform, kanban board, subagents
Tool isolation	no container backends; fail-closed permission engine, approval gates, loopback only	Docker sandbox, off by default; pairing for unknown senders	7 terminal backends (local, Docker, SSH, Singularity, Modal, Daytona, Vercel Sandbox); smart command approval by default
Where it runs	your computer; binds to 127.0.0.1 only	a laptop, or one Gateway shared by a team	a laptop, “a $5 VPS, a GPU cluster, or serverless”
Moving in	imports OpenClaw and Hermes memory files	imports Codex, Claude Code and Hermes memory	hermes claw migrate: OpenClaw settings, memories, skills, API keys
OS	macOS, Windows, Linux	macOS, Linux, Windows; iOS and Android apps	Linux, macOS, WSL2, native Windows, Termux

Sources checked (2026-09-17); competitor columns are as of that date.

OpenClaw — github.com/openclaw/openclaw (README, MIT LICENSE, stars, release v2026.9.4) · channels · memory · MCP · sandboxing · security · cron jobs
Hermes Agent — github.com/NousResearch/hermes-agent (README, MIT, stars) · hermes-agent.nousresearch.com · messaging · memory · MCP · security · kanban
Why people switch — Kilo, 1,300 Reddit comments analyzed · Bito, OpenClaw alternatives · Composio, OpenClaw alternatives

Vodou column checked against the product on 2026-09-17.

Fair to both

When should you choose OpenClaw instead of Vodou?

Choose OpenClaw when free and open matter more than anything else on this page, or when the assistant should live in the chat apps your family or team already use. Vodou is the better fit only if you need the same memory in your browser AI chats, Claude Code or Cursor, and you accept a free account, a monthly price after the trial and a proprietary engine.

You want no cost, no account and an MIT license you can fork — OpenClaw has no paid tier.
You need a channel Vodou has no first-party adapter for, such as Matrix, LINE, WeChat or IRC, or native iOS and Android apps.
You want tools to run in a Docker sandbox, or one Gateway shared by a whole team.
You rely on ClawHub skills and plugins, or on the size of OpenClaw's community when you need help.
You want the assistant on a server rather than your own computer; Vodou binds to 127.0.0.1 and runs where you work.

How it works

How do you move from OpenClaw to Vodou?

Vodou's importer reads an OpenClaw workspace directly: install Vodou, preview with vodou-core mem import openclaw --dry-run, then run it without the flag. MEMORY.md, USER.md and every Markdown note under memory/ are copied into Vodou's local memory and indexed; skills, channel settings, session history and API keys are not moved.

STEP 1
Install Vodou and sign in.
Download Vodou for macOS, Windows or Linux and sign in with a free Vodou account; the 60-day trial starts with no credit card. Add a model key or point Vodou at Ollama. OpenClaw can keep running alongside while you try Vodou.
STEP 2
Preview the import.
Run vodou-core mem import openclaw --dry-run. Vodou looks in ~/.openclaw/workspace, then ~/.openclaw, for a MEMORY.md or a memory/ folder and lists what it would write. If your workspace lives somewhere else, pass its path after the word openclaw.
STEP 3
Import the memory.
Run the same command without --dry-run. MEMORY.md and USER.md land as MEMORY-imported.md and USER-imported.md, so Vodou's own files are never overwritten; every note under memory/ keeps its original date; HTML comments and invisible characters are stripped; lines that look like prompt injection are kept but flagged for your review.
Deep dive: AI memory
STEP 4
Reconnect channels and check the result.
Connect Telegram, Slack, WhatsApp or another channel in Vodou's Messaging settings, then search the imported memory in the Vodou Console. If anything looks wrong, vodou-core mem import-undo <job-id> removes every OpenClaw and Hermes import so you can run it again.

FAQ

Frequently asked questions about OpenClaw alternatives

What is the best OpenClaw alternative?

It depends on why you are leaving. Hermes Agent is the closest open-source swap and most 2026 guides list it first; NanoClaw and ZeroClaw are lighter forks; Claude Cowork is hosted. Vodou fits when the problem is memory reach — the same memory in ChatGPT, Claude Code, Cursor and your chat apps — and you accept a free account and $10 a month on your own keys after a 60-day trial.

Is OpenClaw free?

Yes. OpenClaw is MIT-licensed and its README says it has “no paid tier, hosted service, or token”; you pay only your model provider. Vodou is not free forever: it is free for 60 days with no credit card, then $10 a month on your own model keys, and it requires a free Vodou account. Both can use a local model through Ollama, so model calls can cost nothing.

Is OpenClaw safe?

It can be run safely, but it takes care. OpenClaw runs tools on the host unless you turn on its sandbox, which is off by default; CVE-2026-25253 allowed one-click code execution before v2026.1.29 fixed it; and researchers found malicious skills on ClawHub in early 2026. OpenClaw's security guide covers hardening. Vodou binds only to 127.0.0.1 and gates risky actions behind approvals — any agent you run yourself still needs the same care.

Is OpenClaw open source?

Yes. OpenClaw is open source under the MIT license and stewarded by the OpenClaw Foundation, an independent 501(c)(3). Hermes Agent is MIT-licensed too. Vodou is open-core: its gateway, skills, Vodou Bridge extension and MCP servers are Apache-2.0 on GitHub, while the Rust engine that does routing and memory is proprietary and licensed through a free account.

OpenClaw vs Hermes Agent: which is better?

They make different bets. OpenClaw is a control plane: one Gateway, 32 listed messaging channels, phone apps and shared team deployments. Hermes Agent is a learning loop: it writes and refines its own skills, runs on a $5 VPS or serverless backends, and imports an OpenClaw setup with hermes claw migrate. Vodou imports memory from both and makes that memory usable in your browser, Claude Code and Cursor.

Is there an OpenClaw alternative with memory?

Yes. Hermes Agent keeps a small curated memory plus searchable session history, and several hosted assistants pitch long-term memory. Vodou is memory-first: a local SQLite store with full-text, vector and reranked search, where every fact keeps its source and can be corrected or undone, reachable from 22 browser AI sites, Claude Code, Cursor, MCP clients and your messaging channels.

Is there an OpenClaw alternative for Mac and Windows?

Yes — though OpenClaw itself runs on both, with native apps and a Windows installer. Hermes Agent ships Hermes Desktop for macOS 12+ and Windows 10/11. Vodou installs natively on macOS 13+ (Apple Silicon and Intel), Windows 10/11 x64 and Linux, and runs its memory, gateway and MCP host on your own machine, listening only on 127.0.0.1.

Can I use Claude Code instead of OpenClaw?

For coding, yes. Claude Code is Anthropic's coding agent for the terminal and editor; OpenClaw is an always-on assistant that lives in your chat apps, and it can even run Claude as a harness plugin. Vodou connects the two: its hooks give Claude Code a searched local memory on every prompt, and the same memory answers you in Slack, Telegram or WhatsApp.