Always On Agent — AI skill for Claude Code
Agent Skill for building a personal AI agent that runs continuously on your own machine: four layers to keep separate, waking a laptop that sleeps, approval gates, and a read-only first pass that maps.
How to install Always On Agent
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open Hiberius/always-on-agent and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Always On Agent does
Agent Skill for building a personal AI agent that runs continuously on your own machine: four layers to keep separate, waking a laptop that sleeps, approval gates, and a read-only first pass that maps projects and credentials without ever reading a secret's value.
Alternatives in AI
- Claude Squad — by smtg-ai - Claude Squad is a terminal app that manages multiple Claude Code, Codex (and other local agents i 6.5k ★
- Cwc Long Running Agents — Claude Code's built-in /goal command gives you a generator/evaluator loop out of the box: set a completion con 668 ★
- Obsidian + Claude Code: Rebuild Your Second Brain With AI — Build an AI-powered personal knowledge management system from scratch 540 ★
README
Always-On Agent
**An Agent Skill for building a personal AI agent that runs continuously on your own machine: the four layers to keep separate, how to wake a laptop that sleeps, which phone channel will not get your number banned, and a read-only first pass that maps your projects and credentials without ever reading a secret's value.**
[](LICENSE)   
npx skills add Hiberius/always-on-agent
Phase zero: look before you touch
The first thing an agent does on your machine must be *incapable* of doing harm. Not careful. Incapable.
python3 scripts/survey.py secrets ~/
Key NAMES only. No value in this output was ever read into memory.
/Users/you/work/api/.env (7 keys, 4 sensitive) <- TRACKED BY GIT
STRIPE_SECRET_KEY, DB_PASSWORD, JWT_SECRET, WEBHOOK_SIGNING_KEY
/Users/you/side/bot/.env (3 keys, 2 sensitive)
TELEGRAM_TOKEN, OPENAI_API_KEY
2 secret file(s) found.
1 of them are NOT gitignored inside a repository. Fix that before anything else.
It reads the name on the left of the assignment and discards everything else before returning anything. The test suite plants a canary value in every fixture and asserts it never appears in any output. That is the contract, not a promise.
python3 scripts/survey.py projects ~/
PROJECT BRANCH DIRTY AHEAD DAYS LAST COMMIT
old-client-site main 34 0 412 wip before holiday
scraper feat/retry 6 3 41 handle 429
api main 0 0 2 bump deps
27 repo(s): 4 need attention, 9 have uncommitted or unpushed w
Related Skills
Asynq Agentd
An open-source daemon that keeps your AI coding agents (Claude Code, Codex, OpenCode) running 24/7 on any alwa
ML Brainclone
Scaffold for building a personal AI second brain with Claude Code + Obsidian. Four agents, privacy layers, nig
Cwin
Windows CLI that screenshots, clicks and types into any window without stealing focus, so an AI coding agent c
Jev MCP
TypeSafe AI Jev MCP server for bounded probabilistic decisions: four read-only stdio tools for choices, routin
Pbi Deep
Run a structured four-phase DAX development workflow with hard gates between phases. Phase A: business context
Sdlc Kit
Harness-neutral AI-native SDLC framework — intent grilling, sha256 tamper-evident approval gates, fresh-contex
Related Agents
Dx Navigator
Continuously maintains superb DX for contributors; performs end-to-end maintenance, proposes improvements, and
Org Designer
Head of People. Continuously evaluates team shape. Detects cracks, gaps, bloat, recurring failures. Proposes s
Bh Verifier
Adversarial verifier of Black Heron audit. Runs AFTER all four lenses. Job is to CHALLENGE findings, not add t