AgentIndex icon
AgentIndex
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Submit Tool
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Home/
Workflow Automation/
atomic-agent
atomic-agent logo

atomic-agent

Active·★ 3.5k·MIT·Updated 2026-10-27

Atomic Agent is a local-first general-purpose agent: the control loop and all state, config, memory, sessions, tasks and traces, stay on your machine. From a terminal TUI it browses the web, reads and edits files, runs approved shell commands, inspects documents, remembers context across sessions and schedules follow-ups. It can run open-weight models locally through the team's TurboQuant llama.cpp build, which is tuned so long sessions with small local models fit in less memory, or call cloud models.

Install is a checksum-verified script on macOS, Linux or Windows that drops the CLI (with an atag alias) and support assets; the agent updates itself in place. On first run it offers to import from another agent, tick the sources it finds and it brings over skills, memory, MCP servers and sessions, and opt-in provider keys, from Claude Code, Codex, Oh-My-Pi, Pi, Hermes or OpenClaw without touching the source. A single uninstall command removes everything after printing what it will delete and asking you to confirm.

This is an early build: the README labels it a developer preview whose APIs, commands, config and behavior are still moving, and suggests pinning a release for a stable integration. Current builds cover Apple Silicon Macs, Linux x64/arm64 and Windows x64; Intel Macs are not supported yet.

This write-up is based on the project's README and public repository data. We have not hand-tested the tool. Data as of 2026-10-25.

License
MIT
Latest release
v0.6.11 · 2026-10-24
Last commit
2026-10-25
Contributors
27
Forks
272
Open issues and PRs
99
First published
2026-04-21

Source: GitHub, npm and PyPI public APIs.

#local-first#llama-cpp#local-llm#browser-automation#agent-memory#cli#tui
$ Install
$ curl -fsSL https://atomicagent.io/install | sh
↗ Visit site★ GitHub
01

Features

01Runs local open-weight models on a TurboQuant llama.cpp build, or cloud models
02Browser control, file editing, approved shell commands and document inspection
03Memory across sessions and scheduled follow-ups
04Imports skills, memory, MCP servers and sessions from Claude Code, Codex, Hermes, Pi and OpenClaw
05Checksum-verified installer and a one-command uninstall with dry run
02

Why choose it

+Local-first: the control loop and all state stay on your machine, with a privacy-oriented design.
+Runs small open-weight models efficiently via a TurboQuant llama.cpp build, or cloud models.
+One-click import of skills, memory and sessions from Claude Code, Codex, Hermes and others.
03

Trade-offs

!A developer preview: the README warns APIs, commands and behavior are still moving.
!No Intel Mac support yet; builds are limited to Apple Silicon, Linux and Windows x64.
!Running capable models locally still depends on your hardware.
04

Maintenance check

ActiveChecked: 2026-10-24
Last commit
2026-10-23
Commits, last 90 days
100+
Releases, last 90 days
30
Latest release
v0.6.10 (2026-10-21)
Open issues
36
Recent issues with at least one comment
0/3
Commits per week, last 12 weeks
Commits per week, last 12 weeks: 35, 156, 62, 81, 198, 104, 88, 98, 20, 106, 144, 2

Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.

Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.

Most discussed open issues

  • #229 Feature: User-configurable shell guard rules via config.json + /guard menu command6 comments · 2026-08-22
  • #33 Feature request: Hindsight as memory implementation 6 comments · 2026-07-25
  • #485 feat: add AnySearch as os.web.search provider (+ starter skill)2 comments · 2026-09-23
  • #686 Desktop 0.0.3: agent gets 401 from managed llama-server (core rotates LLAMA_API_KEY, agent reads stale llama-server.key)1 comments · 2026-10-21
  • #602 [Feature / Bug] Windows Subshell & Tooling Gaps: Missing `os.fs.mkdir`, hardcoded `cmd.exe`, and context bloat cause failure loops on local 1 comments · 2026-10-04

Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.

05

Compatibility

macOS (Apple Silicon)
CLI / TUI
Verified via docs
Linux
x64 / arm64
Verified via docs
Windows
x64
Verified via docs
06

Quick start

1
$ curl -fsSL https://atomicagent.io/install | sh
07

Use cases

↳A private assistant that runs entirely on local models
↳Migrating from Claude Code or Hermes while keeping skills and sessions
↳Long sessions with small local models on limited memory
08

Alternatives

⚡Top VS:
VSObserver★1.7k→VSshowagent★63→VSmcp-graph-workflow★41→VSprism-coder★156→
Observer logo
Observer★ 1.7k
Why observe computer if computer can observe for you
VSCompare→
showagent logo
showagent★ 63
Every AI coding session on your machine, in one TUI: browse, search, resume, branch — and convert conversations between Codex, Claude Code and more.
VSCompare→
mcp-graph-workflow logo
mcp-graph-workflow★ 41
MCP local-first CLI tool that converts PRD text files into persistent execution graphs (SQLite) for structured agentic workflows
VSCompare→
prism-coder logo
prism-coder★ 156
The Mind Palace for AI Agents - HIPAA-hardened Cognitive Architecture with on-device LLM (prism-coder:7b), Hebbian learning, ACT-R spreading activation, adversarial evaluation, persistent memory, multi-agent Hivemind and visual dashboard. Zero API keys required.
VSCompare→
kimun logo
kimun★ 64
Simple note taking, powerful search, AI ready
VSCompare→
career-ops logo
career-ops★ 77.9k
Open-source AI job search agent and job finder: scan job boards, score each job 1-5 against your CV before you apply, tailor an ATS-friendly resume and cover letter, get interview prep and a job application tracker. It helps you fill in each application; you press Submit. Runs locally in your AI coding CLI (Claude Code, Codex, OpenCode and more).
VSCompare→
lilbee logo
lilbee★ 66
Chat with your documents offline using your own hardware. Augment any AI agent via MCP with hybrid RAG search over PDFs, code, and 150+ formats. Integrate with your favorite GUI via REST API.
VSCompare→
pinvou-agent logo
pinvou-agent★ 2.7k
Open-source desktop AI agent for tools, files, knowledge, workflows, and real deliverables.
VSCompare→
See all alternatives →

Comments

Log in to leave a comment

No comments yet. Be the first!

On this page
01Features02Why choose it03Trade-offs04Maintenance check05Compatibility06Quick start07Use cases08Alternatives
AtomicBot-ai /
atomic-agent
↗

A local-first agent that keeps its loop and state on your machine, running small open models via a TurboQuant llama.cpp

3.5k272MIT
Last commit1d ago
StatusActive
CategoryWorkflow Automation
Contributors27
Latest releasev0.6.11
Links
Documentation↗Discussion↗Issues↗Releases↗
Cloud Infrastructure
Ad

Deploy on DigitalOcean

Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.

Claim $200 Free Credit
© 2026 AgentIndex.app|Built by a 10-year iOS Developer.
Best by use caseAboutContactAdvertisePrivacy PolicyTerms of UseQYSGitHubBuy me a coffee ☕

Browse by Category

MCP ServersCode AssistantWorkflow AutomationRAG / Knowledge BaseMulti-AgentBrowser AutomationLLM InfraDev ToolingObservability

Not affiliated with Anthropic, OpenAI or Microsoft.