startq 0.3.0

Never lose your AI chats or work again. Boot, monitor, and shut down your AI workflow.

StartQ

Never lose your AI chats or work again.

Your AI Workflow Is a Computer. Boot It Like One.

Think about what happens when you turn on your PC.

The BIOS runs. It checks the hardware. The OS loads drivers, mounts disks, reads your settings, restores your last session. Everything you were working on comes back exactly where you left it. Then you work. Then you shut down — and the OS saves your state so tomorrow picks up clean.

Now think about your AI development workflow.

You have API connections. Database credentials. Multiple AI models — Claude, Gemini, GPT, local models. You have agent scripts, deployment pipelines, live services, project configs across different repos. You have context — decisions made yesterday, bugs found last week, architecture choices from last month.

Every time you start a new AI session, all of that context is gone. You re-explain everything. The agent doesn't remember what you built yesterday. An IDE update wipes your environment. You lose hours.

StartQ treats your AI workflow like a computer. Boot it. Monitor it. Shut it down.

StartQ          AutoQ           EndQ
  │               │               │
  ▼               ▼               ▼
POWER ON       MONITORING      SHUTDOWN
  │               │               │
  ▼               ▼               ▼
Health check    Watch agents    Capture state
Load context    Review work     Git snapshot
Verify auth     Surface issues  Sign receipt
Spawn daemons   Auto-recover    Sync to cloud
  │               │               │
  ▼               ▼               ▼
 Ready          Running         Saved

Three Python scripts. That's it.

The Lifecycle

StartQ — Power On

startq boot

Like pressing the power button on your PC. StartQ runs a boot sequence:

1. POST — checks that your local brain exists and is writable
2. Auth — verifies your identity from config
3. Memory — loads your last session's context, verifies its SHA-256 signature
4. Daemons — spawns any background scripts you've configured
5. Handoff — your AI agent starts with full context from yesterday

You don't re-explain your project. You don't re-paste your API keys. You don't re-describe your architecture. The agent already knows.

AutoQ — Monitor

While you work, AutoQ watches your session:

• Surfaces recurring workflow patterns
• Reviews agent behavior for quality
• Catches drift before it becomes a problem
• Promotes good patterns into reusable workflows

Think of it as Task Manager for your AI workflow. It runs in the background and makes sure things stay on track.

EndQ — Shutdown

startq end -c "Built the payment API. Stripe integration complete. Deploy tomorrow."

Like shutting down your PC properly instead of yanking the power cord:

1. State capture — your summary becomes persistent context
2. Git snapshot — current branch, uncommitted files, all recorded
3. Signing — SHA-256 hash ensures nobody tampers with the receipt
4. Cloud sync — optionally pushes to a hosted brain for multi-machine access
5. Receipt — a signed JSON file in .startq/brain/ proves the session existed

Tomorrow when you run startq boot, all of this loads automatically. Zero amnesia.

Why This Matters — A Real Story

On May 19, 2026, Antigravity IDE pushed an update that wiped environment state for developers mid-session. Workflows deleted. Context gone. Developers lost hours of work.

I didn't lose a single thing.

Because StartQ had already persisted my session state, agent context, and workflow definitions to local memory before the update hit. When the environment came back up, I ran startq boot, and everything was exactly where I left it.

StartQ is insurance for your AI workflow. Your context survives IDE crashes, model swaps, environment resets, and cloud outages.

Install

pip install startq

Zero dependencies. Pure Python standard library.

Quick Start

# Initialize your project brain
startq init

# Start your session (loads previous context)
startq boot

# ... do your work with any AI tool ...

# Save your session (captures context + git state)
startq end -c "What I accomplished today."

Cloud Sync (v0.2.0)

Connect to a hosted brain for multi-machine, multi-agent persistence:

# One-time setup
startq upgrade --url https://your-brain-endpoint.run.app --key sk-xxx

# Now boot and end auto-sync to cloud
startq boot      # pulls latest context from cloud
startq end -c "..." # pushes session to cloud

# Force local-only when needed
startq boot --local

Check Your Status

startq status

Shows your identity, session count, cloud connection status, and latest session context.

What Gets Persisted

Data	Where	Format
Session context	.startq/brain/<uuid>.json	Signed JSON
Git branch + dirty files	Inside session receipt	Embedded
Agent identity	.startq/config.json	JSON
Cloud config	.startq/config.json	JSON
Init state	.startq/state.json	JSON

The PC Boot Analogy

PC Boot	StartQ
BIOS POST	startq boot — health check
Load OS	Load last session context
Mount drives	Connect to cloud brain
Restore session	Inject context into agent
Task Manager	AutoQ — monitor workflows
Save & Shutdown	startq end — sign + persist
Hibernate	Session receipt in .startq/brain/
Wake from sleep	Next startq boot — instant context

Features

• Local First — your brain is a local JSON directory. No cloud required.
• Cryptographic Integrity — every session receipt is SHA-256 signed. Tampering detected on boot.
• Cloud Sync — optionally connect to any REST endpoint for multi-machine persistence.
• Universal — works with Claude, Cursor, Antigravity, Gemini CLI, or any AI tool.
• Daemons — spawn background scripts on boot, like systemd for your AI workflow.
• Git Aware — captures branch and uncommitted files automatically.
• Zero Dependencies — pure Python standard library. No pip packages. No curl. No YAML.

How It Works

These are just Python scripts. No magic. No framework lock-in.

• startq/brain.py — reads and writes signed JSON session files
• startq/cli.py — the command-line interface
• startq/cloud_brain.py — optional REST client for cloud sync

You can read the source in 10 minutes. You can fork it and customize it in 20. That's the point.

License

MIT License. Built in Seattle by Phil Hills.