somora
Run a personal AI team with shared long-term memory and switchable models on your own machine.
- Source repo
- thenaxon/somora_agent
- Stars
- ★ 35
- Last updated
- 4d ago
- License
- MIT
- Primary language
- TypeScript
- FA score
- 58/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platformChatGPT · Codex · Claude Code
- Cost
- Free, no paid service needed
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- An individual who wants separate research, coordination, and execution roles on one computer, with agents able to delegate to one another.
- Not a fit if
- Users who require a native Windows installation path
- Teams unwilling to run a local background service
- Organizations requiring a production-stable rather than actively developed system
- Website
- somora.ai ↗ set on GitHub
- Maintenance
- Commits in the last 30 dayslast commit 2026-11-16latest release v2026.1116.11 (2026-11-16)
- Source review
- 58/100 · Major gaps
What does this agent do, and when should you use it?
somora is a multi-agent server that runs on the user's machine and exposes a terminal TUI, browser desktop, and mobile PWA. Each chat agent can have its own persona, private memory, model preferences, and tool permissions, while stable knowledge can be promoted into a shared Obsidian wiki through background dream cycles. A conversation can move among Claude, ChatGPT, Grok, and OpenAI-compatible backends while retaining its history and tools. The system also includes repository-focused builder agents and tools for files, shell commands, tmux, web access, attachments, sub-agents, scheduled triggers, browser control, and external MCP servers. Configuration, sessions, attachments, and agent memory live under ~/.somora/, while the shared wiki resides in a selected Obsidian vault, keeping the deployment boundary on the user's machine except for explicitly configured model providers and external services.
Requests submitted through somora tui, the Web client, or the mobile PWA enter a per-session queue in somora-server; turns initiated by another agent, a sub-agent, Sentinel, or a voice call use the same dispatch path. The execution layer can invoke claude-cli, codex-cli, grok-cli, or an openai-compatible engine and gives them a shared registry covering memory, dream, wiki, file, exec, tmux, web, agents, skills, sentinel, browser, media, projects, and external MCP tools. Facts noticed during chats can be written to an agent's private Markdown memory. The REM, Deep, and Lucid dream phases respectively turn sessions into memory, promote stable memory into the shared wiki, and review that wiki. Retrieval combines SQLite, sqlite-vec, and FTS5 with separate indexes for private memory, the shared wiki, and read-only vault context. For software work, a kind: builder agent plans, edits, runs, and tests inside a pinned repository; it may read memory but does not write memory or dream.
- An individual who wants separate research, coordination, and execution roles on one computer, with agents able to delegate to one another.
- A user of Claude, ChatGPT, Grok, or local models who wants to change backends during a conversation without losing history or tool context.
- An Obsidian user who wants recurring, stable facts from conversations consolidated into a shared long-term wiki.
- A power user who wants scheduled work, browser interaction, file operations, shell commands, and persistent tmux sessions available to one agent team.
- A developer who wants a long-running repository builder while ordinary chat agents handle discussion, orchestration, and handoffs.
- Someone who wants the same local agents, sessions, queue, and memory available from a terminal, desktop browser, and phone.
Key setup facts
Pulled from the install and usage notes below. The project README has the current values.
- Install / run
curl -fsSL https://somora.ai/install.sh | bashnpm install -g somora && somora setupcurl -fsSL https://claude.ai/install.sh | bash && claude auth login
- Config files
~/.somora/config.yaml- Runtime
- Node.js 22.13+
- Install notes cover
- macOS · Linux
How do you install or deploy this agent?
somora supports Linux and macOS and requires at least one usable LLM backend. The recommended path is to run the installer as a normal user; it checks system tools, supplies missing or outdated Node.js, installs somora, registers the background service, and launches the setup assistant.
curl -fsSL https://somora.ai/install.sh | bashIf Node.js ≥22.13 is already installed, use npm and start the setup assistant directly:
npm install -g somora && somora setupDuring setup, connect at least one backend. A Claude subscription can be authenticated through Claude Code, and a ChatGPT subscription through the bundled Codex login. Local Ollama, LM Studio, vLLM, oMLX, or another OpenAI-compatible service is configured in ~/.somora/config.yaml after initialization.
curl -fsSL https://claude.ai/install.sh | bash && claude auth loginsomora codex logintmux is also mandatory, and Node.js 22.13 is the minimum supported release. The installer may ask before using administrator privileges to add missing system dependencies.
How do you use this agent?
After setup, open the terminal client and send a message to the default agent:
somora tuiFor a manual installation, initialize the local configuration and service before opening the TUI:
somora init
somora server start
somora tuiOnce multiple agents exist, initialize the reporting structure shared with the team:
somora team init --principal "<your name>"The Web client is served at https://<host>.<tailnet>.ts.net:18737/web/, and the mobile PWA is under /mobile/ on the same host. Both connect to the same local server and share its agents, queues, sessions, and memory. Remote Web and mobile access use Tailscale HTTPS; the documented Web deployment uses LAN trust and has no separate authentication layer.
What are this agent's strengths and limitations?
- A conversation can switch among Claude, ChatGPT, Grok, and OpenAI-compatible backends while preserving history and a common tool registry.
- Private memory, a shared Obsidian wiki, and the REM, Deep, and Lucid dream phases provide a concrete pipeline for accumulating long-term knowledge.
- The TUI, browser desktop, and mobile PWA operate against the same server, session queues, agents, and memory.
- Chat agents and builders have distinct operating models; builders add long turns, planning, task tracking, compiler feedback, and a repository-bound workspace.
- Core state remains on the user's machine, and local inference services such as Ollama, LM Studio, and vLLM are supported.
- Only Linux and macOS installation paths are documented, and Node.js ≥22.13, tmux, and at least one model backend are mandatory.
- The repository describes the product as actively developed and open to early testers rather than as a mature stable release.
- HTTPS and browser APIs such as microphone, screen sharing, and clipboard access depend on Tailscale configuration, while the Web client uses LAN trust without separate authentication.
- Grok CLI cannot act as a dream or compaction worker, so those jobs require another configured engine.
- Hosted image and video provider implementations follow published interfaces but have not been tested end to end with live accounts.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| somora This agent | 58 · Major gaps | CLIFree | ★ 35 | 4d ago | TypeScript | ChatGPT · Codex · Claude Code |
| APX v2 Agent Crew | 51 · Major gaps | — | ★ 35 | today | TypeScript | ChatGPT · Codex · Claude Code · Claude.ai · OpenAI API · Claude API |
| NextClaw | 61 · Some gaps | Desktop appFree + model costs | ★ 264 | 1d ago | TypeScript | Codex · Claude Code |
| OpenHuman | 57 · Major gaps | Desktop appFreemium | ★ 43k | 1d ago | Rust | — |
How does FollowAgents rate this agent?
Why each dimension lost points
The security document specifies loopback listening by default, request keys, credential locations, some file and command restrictions, update-check data, and opt-out controls, supporting the data-flow and sensitive-data scores. The installer asks before using administrator rights, and some knowledge promotion requires approval. Deductions: chat agents may have all tools by default, agents are not sandboxed, and general confirmation and rollback for external actions are not established. Daily telemetry is enabled by default, and the packet shows no lockfile or dependency audit evidence. Repository attribution is traceable, but publisher identity is unknown and the copyright holder is named “suspect,” so source attribution is only partial.
The README, package metadata, and security document are broadly consistent about Node requirements, platforms, storage, and main engines, and identify required and optional components. Deductions: the packet shows dependency ranges but no resolved lockfile; many product capabilities are asserted in the README without matching implementation or tests in the sample; failure guidance appears in particular setup or configuration contexts, with limited demonstrated coverage.
The README targets concrete scenarios including personal self-hosting, multi-agent collaboration, coding builders, and several clients. It distinguishes chat agents from builders and describes optional features and several environment requirements. Deductions: tool permissions are configurable per agent but the default is broad; trigger applicability is described mainly in the README and sample skill, with limited evidence of wider boundary handling; supported environments are limited to macOS and specified glibc-based Linux.
The documentation is clearly organized, and the README links to quickstart, manual installation, models, clients, and feature guides. Package name, version, repository address, and MIT license are explicit. Deductions: the supplied sample does not include the contributor or security policy contents, an FAQ, a consolidated known-limitations section, or evidence of a named maintainer; release notes are linked but their contents are absent. The skill templates and examples are concrete but do not cover the whole product.
The README describes a shared tool surface, multiple clients, long-term memory, and a multi-model agent team, which would offer clear utility if implemented as stated. The orchestration skill also gives a concrete handoff format and phase choices. Deductions: the supplied sample provides little implementation, runtime, or test corroboration for these outcomes; runtime, disk, and model-service requirements are substantial, so the cost-benefit case is not fully established.
Installation requirements, security boundaries, telemetry fields, and skill setup steps are stated concretely and trace to the supplied files; the README and package.json partly corroborate package version and Node requirements. Deductions: most feature claims lack a second source or code evidence in the sample, and execution outcomes cannot be confirmed from this review. Documentation distinguishes defaults, configuration options, and limitations to some extent, but unverified promotional statements remain claims.
- The security document says agents are not sandboxed; shell and file tools run with the service user's permissions. Grant chat-agent tools carefully and protect the server key.
- The daily version check is enabled by default and sends version, OS, Node version, and CPU information; it can be disabled with DO_NOT_TRACK=1 or configuration.
- The installation instructions pipe a remotely fetched script directly to bash; users should review the installer and its download source.
- This assessment uses only the supplied static sample and cannot confirm README feature claims, test status, or runtime behavior.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
v2026.1116.11 v2026.1116.11 — coding-agent teams in tmux, a start menu, your answered questions stay in the chat
Release notes excerpt
Before you update Nothing to do by hand. What you may notice: - The tmux tool reports agent: { name, state } instead of tui_state. The state is one of idle, working, blocked (a dialog waits for an answer) or unknown, read with [herdr](https://github.com/herdrdev/herdr)'s agent rules. kind is no longer needed: the agent is recognised by its process. An agent's own notes or skills that look for tui_state should look at agent.state. …View on GitHub ↗ -
v2026.1111.1 v2026.1111.1 — leaner builders, a security review, provider tools kept out, every dependency current
Release notes excerpt
Before you update Nothing to do by hand. What you may notice: - Builders work differently. A builder's rules are now short tool rules instead of a method (tests as proof, evidence for every claim, a task list kept every few rounds); measured on five tasks against another coding harness on the same model, the short rules did every task at least as well and the hardest one better. A new session starts building right away — switch on the plan phase in the task panel when you want a plan first. The reminders to update the task list are gone; the list stays. Chat agents are unchanged. …View on GitHub ↗ -
v2026.1106.3 v2026.1106.3 — a settings window for config.yaml, builders held to what they deliver, voice calls from the phone and on your own models
Release notes excerpt
Before you update Nothing to do by hand. What you may notice: - Builders work longer and are checked more. A builder is no longer told to wrap up at 75 % of its budget: it gets a checkpoint there and keeps working, and it has no round cap by default (the 2000 tool calls, the 8-hour clock and the loop guard brake a build). An order from another agent comes with a required acceptance list, and under every builder report somora adds the state of the working copy as it sees it. Chat agents are unchanged. …View on GitHub ↗
FAQ
Is a paid model subscription or cloud API required?
Where is data stored by default?
~/.somora/. Workspace files and generated media go to the configured workspace, while the shared wiki resides in a selected Obsidian vault. Exposure to external models or services depends on the user's configuration.Are all capabilities enabled immediately after installation?
imageGen or videoGen block is present.How does a builder differ from a chat agent?
What should I consider before exposing the Web client remotely?
Is somora open source?
Is somora safe to use?
Is somora still maintained?
Who makes somora?
What is somora's official website?
Thanks — we use this to improve the page.