OtoDock

Why STEADY

STEADY (73) for a coherent, ambitious self-hosted agent platform with an unusually concrete and correct security posture (kernel sandbox per agent, network isolation of private/metadata ranges, per-session credential injection, SSO/2FA/budgets), honest Fair Source licensing, and a genuine free self-host tier up to five users. Held in the middle of the band because it is a public-surface review (medium confidence, not self-hosted or exercised by Hlido), the feature breadth is very large and almost all of it is unverified from the surface, adoption requires Docker self-hosting plus a customer's own Claude/Codex subscription, and community traction is modest (187 GitHub stars).

Public-surface checklist

• PASS Homepage loads (required) — otodock.io returned a full product page (page_title 'OtoDock — Collaborative Agents')
• PASS Primary value prop (required) — 'The agentic company OS' — a self-hosted team of AI agents on infrastructure you control, built on Claude Code & Codex
• PASS Cta present (required) — 'Get started' / 'View the source' / a self-host install snippet (curl the install.sh, bash install.sh)
• PASS Pricing or access — Free to self-host up to five users; seat licensing for growing teams (pricing page linked; prices exclude VAT)
• PASS Evidence or demo — A two-minute dashboard video and detailed feature/security sections shown; features are illustrated but not independently exercised from the public surface

What it does well

• Coherent, editable agent model — persona, memory, workspace, knowledge, skills and tools are each first-class and user-editable, with clear roles and four personal/shared sharing modes
• Specific, correct security posture for autonomous agents — a kernel sandbox per agent (own mount/process namespaces), always-on network isolation of private ranges/LAN/cloud-metadata endpoints, credentials encrypted at rest and injected per session, SSO/2FA and per-user cost budgets from install
• Runs on the customer's own infrastructure and their own Claude Code / Codex subscription (or API keys) — data and keys stay on hardware they control
• Honest, transparent licensing — Fair Source with the full source public and each release converting to Apache 2.0 after two years, plus a real free tier up to five users
• Low-friction self-host story — a single install script that checks Docker, writes the .env, and starts server, dashboard, PostgreSQL and live document preview

What it fails at

• Very broad feature surface (departments, delegation, live dashboards, remote-machine control, a phone line, in-chat Excel/Word/PowerPoint editing) that is almost entirely unverifiable from the public surface
• Marketing flourish that outruns evidence — 'this entire video was directed, captured and edited by an OtoDock agent' is precisely the claim that needs a hands-on run to substantiate
• Adoption friction — requires Docker self-hosting and a customer's own paid Claude/Codex subscription before any value is realised
• Modest community traction (187 GitHub stars) relative to the scope of what is promised
• No hands-on run by Hlido — agent reliability, sandbox enforcement, and the automation/phone/remote-machine features are asserted by the vendor, not confirmed here

Pricing & access

• ModelOpen source · Free tier · Subscription
• Free entry pointYes — a free tier or open-source edition is documented
• Pricing findable on the public surfacePASS Free to self-host up to five users; seat licensing for growing teams (pricing page linked; prices exclude VAT) (tested 2026-09-28)

Derived from Hlido-held evidence only (engine checklist + editorial text); quotes are verbatim from the scorecard; not vendor-supplied; re-derived daily. Verify current prices on the vendor's pricing page. Last verified 2026-09-28.

Related agents

• Activepieces Workflow & Automation · 90
• Dust Workflow & Automation · 90
• Integrately Workflow & Automation · 90

Agent relevance

Webhook

OtoDock is a self-hosted platform that humans use to build and run agents on Claude Code / Codex backends; it consumes MCP tools from a community catalog to extend its own agents, and supports webhook-triggered automations. It exposes no documented external API or MCP server for a third-party agent to drive OtoDock itself from the public surface — it is agent infrastructure you host and operate, not a component another agent calls remotely. Behavioral testing requires self-hosting behind a login.

Agent-friendly score: 4/10

Evidence

• Self-hosted, multi-tenant 'agentic company OS' built on Claude Code and Codex, running on the customer's own Anthropic/OpenAI subscription; free to self-host up to five users, no credit card — source (2026-09-28) verified
• An agent is composed of six editable parts (persona, memory, workspace, knowledge, skills, tools), with defined roles and four personal/shared workspace modes — source (2026-09-28) verified
• Each server-side agent runs in a kernel sandbox with its own mount/process namespaces and always-on network isolation; private ranges, LAN and cloud-metadata endpoints are unreachable by default; credentials are encrypted at rest and injected per session — source (2026-09-28) verified
• Fair Source licensing — full source public, free self-host up to five users, seat licensing above, and each release converts to Apache 2.0 after two years — source (2026-09-28) verified
• Agents can run automations (recurring, event-triggered, one-off), be given a phone number via Twilio/Asterisk, control remote machines, and edit Office documents in chat — source (2026-09-28)