惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

MyScale Blog
MyScale Blog
J
Java Code Geeks
Vercel News
Vercel News
A
About on SuperTechFans
G
Google Developers Blog
C
Check Point Blog
腾讯CDC
N
Netflix TechBlog - Medium
博客园 - 司徒正美
S
SegmentFault 最新的问题
D
DataBreaches.Net
博客园_首页
美团技术团队
Stack Overflow Blog
Stack Overflow Blog
博客园 - 聂微东
量子位
雷峰网
雷峰网
IT之家
IT之家
小众软件
小众软件
Blog — PlanetScale
Blog — PlanetScale
博客园 - 三生石上(FineUI控件)
H
Help Net Security
宝玉的分享
宝玉的分享
博客园 - 叶小钗

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
Running Hermes Agent on Android: A Production AI Agent in...
Alexandre La · 2026-05-16 · via DEV Community

No server. No cloud bill. Just a phone, Termux, and an AI agent that ships code.

When I first heard about Hermes Agent, the open-source agentic framework from Nous Research, I did what any reasonable developer would do: I tried installing it on my laptop. It worked. Then I asked myself — what if I could run it on my phone, always on, always ready?

Turns out you can. And it's surprisingly practical.

The Setup: Android + Termux

Termux gives you a functional Linux environment on Android. With it, you get bash, Python, Node.js, git — everything you need. Installing Hermes Agent is a one-liner:

pkg install python git
pip install hermes-agent
hermes setup

Enter fullscreen mode Exit fullscreen mode

The setup wizard walks you through picking a provider (I use DeepSeek), configuring tools, and connecting messaging platforms. Within 10 minutes, I had a fully operational AI agent running on a Samsung phone.

Why Run an Agent on Your Phone?

Three reasons:

  1. Always-on. Your phone is already on 24/7. No separate device, no VPS, no cloud bill.
  2. Messaging-native. Hermes Agent connects to Telegram natively — you interact with it like a contact. Wake up, check your messages, ask your agent what happened overnight.
  3. Capable. Despite running on ARM64 with limited RAM, Hermes performs real agentic work: multi-step reasoning, tool orchestration, file manipulation, git operations, web search, and cron scheduling.

What My Agent Does Every Day

Here's what's running right now, autonomously:

  • Portfolio deployment — edits HTML/CSS, commits to git, pushes to GitHub Pages, all triggered by a Telegram message
  • Bounty watching — two cron jobs scan Superteam Earn and GitHub for freelance bounties every 3-6 hours, filter by relevance, and deliver results straight to chat
  • Code generation & file ops — writes scripts, patches files, searches codebases, executes Python — all from my pocket

The agent doesn't just respond to commands. It plans: when I say "update the portfolio footer with my name and remove contact links," it navigates the repo, finds the right file, makes targeted patches, commits, and deploys — in one pass.

The Tools That Make It Work

Tool What it does
terminal Shell commands — git, curl, pip, builds
patch Semantic find-and-replace in files
write_file Create or overwrite files
read_file Read with line numbers, pagination
search_files Ripgrep-powered code search
cronjob Schedule recurring autonomous tasks
send_message Deliver results to Telegram
memory Persistent facts across sessions

Each tool is designed for agents, not humans — the patch tool has 9 fuzzy-matching strategies, read_file paginates, cronjob handles delivery destinations. The ergonomics are right.

Cron: The Secret Weapon

Hermes Agent's cron system deserves special attention. You define a job once, and it runs on schedule — fully autonomous, no human in the loop:

hermes cron add "every 6h" \
  --script ~/.hermes/scripts/superteam_bounties.sh \
  --no-agent \
  --deliver origin

Enter fullscreen mode Exit fullscreen mode

The --no-agent flag means the script runs directly (zero LLM tokens) and its stdout is delivered verbatim. Silent when there's nothing to report.

Real Example: From Message to Deploy in 30 Seconds

Here's a compressed look at an actual interaction:

Me: Remplace bot par IA agentique professionnelle dans le portfolio

Hermes: Fait. C'est en ligne.

The agent had persistent knowledge of the repo structure from memory, found the exact string, applied the patch, committed with a meaningful message, and pushed to main. GitHub Pages deployed automatically.

Lessons Learned

  1. Memory is underrated. Hermes Agent's persistent memory means it remembers your preferences, project structure, and past fixes.
  2. Skills are living documentation. When you solve a complex workflow, save it as a skill.
  3. No-agent crons are efficient. For repetitive data collection, skip the LLM entirely.
  4. The phone is not a toy. An Android phone becomes a production-grade AI workstation.

Try It Yourself

pkg install python git -y
pip install hermes-agent
hermes setup

Enter fullscreen mode Exit fullscreen mode

Connect a Telegram bot, schedule a cron, send a message — you'll have an AI co-worker in your pocket within the hour.


Hermes Agent is open source. Runs anywhere you can install Python — including the phone in your hand right now.