Powering a life operating system.
What Siri wishes it was
Behalf.bot is an opinionated agentic Life OS that bottles productivity, wellness, and business wisdom, and works 24/7 on your behalf.
It ships in two variants. Behalf.bot runs on Claude Code, so you get everything Claude can do at a fixed monthly fee under your own Claude Max subscription. Behalf.bot Open runs on OpenCode, so you can point it at any model you like, open-weight models included, for a stack that is self-sovereign end to end.
This is what Apple Siri wishes it could be. Behalf.bot is always on agent with an opinionated life OS, second brain, tool access, context caching and all the surfaces integrated that you use to enable fluid real-time comms as you move throughout your day. It enables you to rise up to act as captain of your ship instead of a deckhand.
Oh captain, my captain
If you're like most people these days 90% of your time is monopolized by admin and tasks that could be offloaded to someone if only you had a trusty EA with all the necessary access and process documentation to understand how to perform tasks on your behalf. Behalf.bot is the EA you've wanted.
Behalf.bot allows you to extricate yourself from the daily muck of tedious work. It's a hyper-capable agent with two variants offering you the flexibility to power it either with your Claude Max subscription optimizing for a fixed monthly subscription cost and access to the powerful Opus and Fable models or via OpenCode affording you an entirely vendor-agnostic, open source stack powered by open-weight, fully-private and/or on-prem models.
Win back your day.
What if you could confidently offload all the life and business admin that eats up your time to a capable agent that knows how you operate? Behalf.bot is the agent stack that comes bundled with a Life OS which bakes in a handful of productivity frameworks like GTD, One Thing, Tiago Forte's "Second Brain" PARA methodology, Lean Startup, Body for Life, 30x500, Stacking The Bricks, Extreme Ownership, Turn the Ship Around, Design-first development, and The Art of Living giving you the benefits of all of these methodologies inherently without having to go learn them.
You direct Behalf.bot with intent, goals, resources, constraints and judgement calls. The OS interprets it. The agent ships email drafts, PRs, briefings and makes progress overnight while you sleep. You wake up to work already done, queued for your review. You set the guardrails, the level of autonomy you're comfortable with and the interaction model (shared accounts vs. own accounts for the agent) and it honors your preferences.
Introducing Behalfbot.
Behalf.bot runs as an assistant controlled via your favorite messaging platform and running heartbeats to wake up and do things proactively on your behalf. It triages your inbox, drafts customer support replies, opens PRs against your repos, monitors your health data, manages your calendar, interfaces with your CRM and POS systems, runs your social media and communicates summaries so you can command it like a CEO giving directives instead of getting spun out in the weeds of daily operational muck.
Sean Tierney built and open-sourced this project out of frustration with Open Claw and its unreliability for delivering meaningful, consistent work. He has been using it for months now shipping daily refinements to the chassis and recently ported the Claude-centric codebase to support OpenCode as an open source harness alternative. Both variants are free to self-install, and governed by the OSaaSY dev license. And if you prefer a "done for you" option for €999 you can get a hands-off managed install that lets you run it wherever you like and just get the keys to a working setup.
Source at scrollinondubs/behalfbot and scrollinondubs/BehalfBot-open.
A non-exhaustive list of what it does.
As you read this, every system below is running in production against real data. None of it is theoretical.
Customer support and billing
Drafts replies overnight against the tone and policies you set. Surfaces escalations. Reconciles billing edge cases.
Health and habits
Tracks macros and workouts via meal photos and a hand-written journal. Reconciles with Apple Health, Oura, and whatever else you wear.
Infrastructure monitoring
Watches your cloud surfaces (Google Cloud, Vercel, n8n, and a dozen others) on a heartbeat. Alerts only when something needs you.
Daily morning briefing
A custom feed assembled from your calendar, inboxes, repos, and reading list. Everything you need to start the day, none of the noise.
Pacman skill ingestion
Feed it a URL or a GitHub project. It digests the content, evaluates fit, and proposes new capabilities for itself - extending what it can do, gated on your approval.
Code review and shipping
Reviews GitHub PRs against any repo you authorize. Ships fixes against your existing branches. Tests, commits, opens PRs for your review.
Content workshopping
Turns notes and voice memos into blog stub drafts in your voice. Drafts the social media promotion to go with them.
Multi-modal surfaces
Ingests from your reMarkable tablet, voice notes, and Loom recordings. Books indexed for semantic search. Briefings drafted. Complex app behavior modifications explained verbally. Natural language goes in, work product comes out.
Integrated with your Second Brain
Integrates with the top PKM's like Notion, Obsidian and SiYuan for reading and writing structured markdown data. Uses Memory MCP for short-term context persistence and your PKM system for long-term persistence.
Midnight Oil
Ships features while you sleep. A queued backlog of pre-approved work executes overnight; you wake up to PRs ready to merge.
See it in action.
Sean's May 19, 2026 talk unveiling Behalfbot at Vibecode Lisboa - walks through the architecture, the live install on his own systems, and a Q&A on what the chassis can do.
Two ways to get it running.
Same chassis underneath both. Pick the one that matches how much time you want to spend at the terminal.
Self-install
Works the same for either variant. Clone the chassis from GitHub, populate your install-profile.md + chassis.config.yaml, run the bootstrap. Hand-author both files from the in-repo samples and you are looking at ~2-4 hours from cold if you are comfortable with Docker.
Simplify your install by using our online interview process - we generate both files for you. Answer 13 questions, download the bundle. ~30-45 min once your accounts are provisioned.
Run the interviewManaged install
A master Behalf.bot instance drives the install end-to-end over SSH, either variant. From cold host to first heartbeat fire in 60-120 min. No terminal time on your side beyond the initial homework.
- ✓Maintainer-driven SSH install
- ✓Discord channel access during install window until it's proven to be working and useful
- ✓One-month free of VibecodeLisboa.com subscription to ramp up quickly with productivity on your Behalf.bot instance
- ✓Real-time chat support during setup for answering install and usage questions and training on best practices
Early-bird beta cohort
Buy a managed install and you get a free month of Vibecode Lisboa. See what that includes.
Both paths produce the same install - same chassis underneath. Self-install gives you full control over the timeline + cost. Managed install gives you a maintainer who drives the bootstrap and stays available post-install for questions.
How Behalfbot compares.
Named-competitor matrix. Tap any row for the longer read.
+Model costBehalf.bot: flat fee, your Claude Max subscription
Behalf.bot Open: metered, whatever your LLM provider chargesBYO-key, meteredBYO-key, meteredOperator brings modelsOAuth subscription (Claude / OpenAI / Gemini etc)Claude Max subscription
Behalf.bot Open: metered, whatever your LLM provider charges
+OriginMonths of daily operator use, distilledSingle author, experimentalActive, plugin ecosystemAlpha, ~14 starsSingle author, Rust, ~7.1k starsBuilt by Anthropic
+SurfaceDiscord, email, health, PRs, calendar, content - whole lifeDiscord + TelegramTelegram, Discord, Slack, WhatsApp, TeamsTerminal, browser, phoneTerminalDesktop app, mobile app
+CapabilityAnything you can do via terminalLimited tool surfacePlugin-extendedOrchestrates other harnessesCoding-focusedSeverely restricted agentic
+InstallSelf-install free, €999 managedSelf-install onlySelf-install onlySelf-install onlySelf-install onlyDownload the app
+LicenseOpen source (O'Saasy)MITOpen sourceApache-2.0MITClosed
+PhilosophyOpinionated life OSMinimal runtimeMature general harnessMeta control-planeSingle-binary runtimeAgentic capabilities bolted onto app, restricted
Model cost+
Behalf.bot Open: metered, whatever your LLM provider charges
Behalf.bot rides your existing Claude Max subscription, so the model cost is a flat fee no matter how hard you work it. Behalf.bot Open points at whichever provider you choose, so the model cost is metered at that provider's rates - or effectively zero if you self-host open weights. The other harnesses bill metered API tokens or require you to bring your own provider keys. Claude Cowork also runs on Claude Max but the agentic surface inside the app is heavily restricted.
Origin+
Behalfbot emerged from months of Sean running it daily as his own assistant before it was packaged into a chassis. Most of the others were designed top-down and released for adoption. Lindy effect: it does what it does because real use selected for what works.
Surface+
Behalfbot is multi-surface - inbox triage, support drafts, PR review, calendar, health, content stubs, briefings. jcode is coding-focused, Hermes is platform-coverage-focused, Claude Cowork is constrained to its own desktop and mobile apps. Life OS scope is the differentiator.
Capability+
Anything you can do via terminal you can do via Behalfbot. Claude Cowork bolts an agentic loop onto the desktop / mobile app but the surface it can touch is severely restricted - no shell, no arbitrary integrations. The bot / harness category here is broad - we are not.
Install+
Behalfbot is the only one with a managed install option. Free self-install for everyone on either variant, €999 hands-off if you would rather not touch a terminal. Claude Cowork is the easiest install of the bunch - it is a packaged app - but you trade install pain for capability ceiling.
License+
Both variants are open source under the O'Saasy License - free to install and run, with SaaS-rights reservation. See FAQ for details. Most competitors are MIT / Apache; Claude Cowork is closed-source Anthropic product.
Philosophy+
We are an opinionated life OS, not a generic harness. The agent enforces the operator's preferences end-to-end. The other systems are infrastructure you assemble into something. Claude Cowork bolts agentic capability onto a packaged Anthropic app, then locks down what the agent can reach.
Open source. Evolved from months of real daily use. Behalf.bot is the only harness of its caliber that runs legally under your Claude Max subscription for a fixed monthly cost. And Behalf.bot Open is the only fully-open, self-sovereign stack running open-weight models at this level of harness capability.
Common questions.
What is the difference between Behalf.bot and Behalf.bot Open?+
Same operating system, different engine underneath. Behalf.bot runs on Claude Code, so it rides your Claude Max subscription at a fixed monthly fee and inherits every capability Anthropic ships into Claude. Behalf.bot Open runs on OpenCode, so you point it at whatever model you want - a frontier API, a local open-weight model, anything in between - and the whole stack is yours end to end. Pick Behalf.bot for the strongest agent at a predictable price. Pick Behalf.bot Open if you want no vendor in the loop at all.
Is the chassis really open-source?+
Yes. Both variants. Behalf.bot at github.com/scrollinondubs/behalfbot and Behalf.bot Open at github.com/scrollinondubs/BehalfBot-open, both under the O'Saasy License (MIT-shaped with light attribution and a SaaS-rights reservation). Clone it, modify it, ship it. No keys required, no behalf.bot account needed - signup only matters if you want the interview to generate your install bundle for you, or you want the managed-install path.
How does it run on my Claude Max subscription?+
That is the Behalf.bot variant. It uses your existing Claude Max OAuth session to talk to Anthropic - the same subscription you already pay for. No metered API tokens, no per-call billing, no surprises on the invoice. The harder you work it, the more value you extract from a fixed bill. If Anthropic ships a new feature into Claude Max, your bot inherits it for free. Behalf.bot Open does not use Claude Max at all; it bills through whichever provider you configure.
Is this just another wrapper on top of an LLM?+
No. Behalfbot is a long-running agentic harness with its own memory layer, scheduled-task system, ops dashboards, audit trails, and skill-extension pipeline. The LLM is one component, and on Behalf.bot Open it is a component you swap. The infrastructure that lets the agent operate continuously and safely against your real surfaces is the rest of it.
What does it cost to run?+
On Behalf.bot: your existing Claude Max subscription plus a small amount for the third-party APIs the agent uses (OpenAI for embeddings, Notion, etc.) - around $20/mo extra in Sean's setup. On Behalf.bot Open: whatever your chosen model provider meters you, which can be close to nothing if you run open weights on your own hardware. The chassis itself is free either way. The only line item from us is the optional one-time €999 managed install fee. No subscription. No per-seat charge.
Where does my data live?+
On infrastructure you control. The agent runs on your host (Linux server, Mac, or VPS); your second brain lives in your Notion / Obsidian / SiYuan; your secrets in your password manager. Nothing phones home to behalf.bot. You see every prompt, every tool call, every decision in your local logs. Full audit trail. On Behalf.bot Open with a locally hosted model, nothing leaves your machine at all.
What can it actually decide on its own?+
Whatever you authorize, and nothing more. The default install ships with a tight allowlist - read everything, approve before writing, never spend money, never send external email without explicit confirmation. You expand the autonomy envelope as you build trust.
What's grandfathered pricing?+
The price you buy at is the price you keep. Buy a managed install today at €999 and future installs on the same primary email stay at €999, even if we raise the price for new customers. Earlier customers keep the rate they bought at - a price change never touches an existing customer. Right now every managed install purchase also includes one free month of VCL Premium at Vibecode Lisboa (normally €199/month).
Can I host it for someone else?+
Hosting a single Behalfbot install for one person or team you support directly is fine - that's how the managed-install path works, and that's what the license permits. What the O'Saasy license reserves for the original maintainer (Vibecode Lisboa) is the right to offer Behalfbot as a hosted, managed, or Software-as-a-Service product to third parties where the chassis itself is the value being sold. Those commercial-SaaS rights stay with us, with no time limit. If you're thinking of running a Behalfbot-as-a-Service business on top of the chassis, talk to us first - the SaaS opportunity here isn't open by default.
Get a force multiplier that allows you to do big things.
Self-install either variant for free, with or without the interview. Or take the managed install (€999, hands-off).
Every managed install includes a free month of Vibecode Lisboa.
Want the interview to generate your config files? Run the interview.