#1111th of 54 in Agent and MCP starters
mercury-agent
Personal agent daemon with permissioned tools, local memory and chat-app access
Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 3.9k
- License
- MIT
- Last commit
- Nov 2026
- Last release
- Nov 2026
- Language
- TypeScript
340 on 21 Apr 20263,896 on 10 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Mercury Agent is a TypeScript agent that runs as one background process on your machine and is reachable from a terminal UI, Telegram, Discord, Slack, Signal and a local web dashboard. It has about 50 built-in tools (files, shell, git, GitHub, fetch, scheduler), a per-folder permission and approval layer, and SQLite-based memory. It also includes a coding mode and persistent scheduled bots.
Who it is for: Self-hosters wanting an always-on personal agent reachable from chat apps
Strengths
- Permission layer: shell blocklist, per-folder scopes, approval prompts and SSRF-guarded fetches
- One daemon shared by all channels, with crash recovery and a login service on each OS
- Memory stays local in SQLite with FTS5; daily token budget enforced per turn
- Many providers with automatic fallback, including Ollama and LM Studio for local models
Weaknesses
- No Docker or Compose setup; installs as a host binary or npm package
- Community skills are not audited; the README tells users to review them first
- Release binaries are unsigned; macOS builds need your own codesign and notarization
- Built as a personal single-owner agent, not a multi-tenant backend to fork
What it needs
- TypeScript, DeepSeek, OpenAI, Anthropic, Grok
- Needs Node.js 22+ (npm install only), signal-cli (Signal channel), LLM provider API key or local Ollama/LM Studio
- env example file
Quick answers
Is mercury-agent open source?
Yes. Its code is on GitHub under the MIT license.
Also in Agent and MCP starters
See all 54| Rank | Project | Score |
|---|---|---|
| 1 | deepseek-harnessPlugin-based agent harness from DeepSeek with a local Web UI | 98 out of 100 |
| 2 | composioTypeScript and Python SDKs that give agents authenticated tools across 1000+ apps | 92 out of 100 |
| 3 | conductorDurable workflow engine that orchestrates microservices and LLM agents from JSON | 91 out of 100 |
| #4 | mcp-useTypeScript framework for building MCP servers with React Views and an Inspector | 87 out of 100 |
| #5 | a2uiProtocol and renderers for agents to stream native UI instead of text | 86 out of 100 |