aeon is the most autonomous agent framework: open source and MIT licensed. You give it a direction and it leverages 78 skills to get the work done, running unattended on GitHub Actions with self-healing loops, reactive triggers, and persistent memory.

Is aeon a hosted API or SaaS?

No. aeon is a framework you fork and run on your own GitHub Actions. There is no public REST endpoint to call - the unit of integration is a skill (a SKILL.md with a schedule and when-to-use guidance).

How do I get started with aeon?

Paste one line into any coding agent (Claude Code, Codex, Cursor): "read https://www.aeon.fun/skills/aeon.md and follow the instructions to set up your aeon agent". It forks aeon, authenticates a model, wires a notification channel, and schedules the first skill.

How much does aeon cost?

aeon itself is free and open source under the MIT license. You pay only for what it runs on: GitHub Actions minutes and your own model API usage (an Anthropic/Claude key to start).

Is aeon affiliated with AEON payments or aeon-toolkit?

No. aeon (the agent framework at aeon.fun) is not affiliated with AEON, the agentic-economy payments network, or with aeon-toolkit, the Python time-series machine-learning library.

Aeon vs ChatGPT: the agent that works while you're gone

ChatGPT answers when you ask. Aeon runs when you don't. Not about which chatbot is smarter, but where a chatbot stops being enough.

PUBLISHED AUG 25 2026READ 4 MIN

Aeon vs the Others, #3. ChatGPT answers when you ask; Aeon runs when you don't.

ChatGPT is the best-known way to talk to an AI. Open the app, type a question, get an answer, close the app. Aeon is an open-source agent framework that runs unattended on GitHub Actions. You configure it once, pick skills, set a schedule, and it keeps shipping work (briefs, monitoring, code, research, disclosures) with your laptop closed.

So this page is not about which chatbot is smarter. They do different jobs, and the useful question is where a chatbot stops being enough.

Set up Aeon. Free, MIT-licensed, runs on your own GitHub account.

The two in one paragraph each

Aeon is a framework you fork and run. A skill is a Markdown file with some frontmatter and a prompt, scheduled on cron and executed in the cloud by any of ten agent harnesses (Claude, Grok, Codex, Pi, Vibe, Kimi, fx, Cursor, Hermes, GLM). A model scores every run 1-5, and failing skills get diagnosed and patched by a self-healing loop. Results land in your Telegram, Discord, Slack, Buzz, or inbox.

ChatGPT is OpenAI's general-purpose assistant, used through its own web and mobile apps. Great at conversation, writing, analysis, one-off tasks. It has scheduled Tasks and some agent features, but the product assumes you're there: you open the app, you drive, you read the output in the app.

Side by side

Runs without you (machine off, app closed) - Aeon: GitHub Actions, cloud-native. ChatGPT: built around live sessions.
Cron scheduling - Aeon: native, per-skill in aeon.yml. ChatGPT: Tasks exist, limited scope.
Reacts to conditions (reactive triggers) - Aeon: schedule "reactive". ChatGPT: none.
Self-healing (auto-diagnoses and patches failures) - Aeon: skill-health plus skill-repair. ChatGPT: none.
Quality scoring on every run - Aeon: model-scored 1-5. ChatGPT: none.
Persistent memory - Aeon: file-based, version-controlled (SOUL.md, STRATEGY.md). ChatGPT: opaque, app-managed memory.
Ships code to your repos - Aeon: PRs, reviews, vuln disclosures. ChatGPT: writes code in chat, you apply it.
Delivers to your chat apps - Aeon: Telegram, Discord, Slack, Buzz, email. ChatGPT: its own app.
Model choice - Aeon: 10 harnesses (Claude, Grok, Codex, Pi, Vibe, Kimi, fx, Cursor, Hermes, GLM). ChatGPT: OpenAI models.
Open source - Aeon: MIT. ChatGPT: no.
Fleet of specialized instances - Aeon: spawn-instance, isolated billing. ChatGPT: no.
Talking to it casually on your phone - Aeon: via Telegram commands. ChatGPT: the app is very good.
Zero setup - Aeon: fork a repo, add secrets. ChatGPT: sign up and type.

Pricing

Aeon: free. MIT license, and Actions minutes cost nothing on public repos. You bring your own model auth, either a Claude subscription or API keys for whichever harness you run.

ChatGPT: Free tier, Plus $20/mo, Pro $200/mo, Business per seat.

In practice: ChatGPT's price buys you a better session. Aeon's cost (your model usage) buys output that piles up while you sleep.

Pick ChatGPT if

You want answers, not automation. Questions, drafts, explanations, images.
You want zero setup and a good mobile app.
Your work is conversational, so you're in the loop anyway.

Pick Aeon if

The work repeats: daily briefs, market monitoring, repo watching, PR review, research digests.
You want results pushed to Telegram/Discord/email instead of pulled from an app.
You want to pick the underlying model, or swap it later, instead of marrying one vendor.
You want the agent to notice its own failures and fix them.
You'd rather own the whole stack. Skills are Markdown files in a repo you control.

Running both is normal. ChatGPT for the conversation, Aeon for the stuff that shouldn't need one.

FAQ

What is Aeon and how does it run without my computer?

Aeon lives in a GitHub repo you fork. Skills run as GitHub Actions workflows on a cron schedule, so execution happens on GitHub's infrastructure. Your laptop can be closed, off, or in another country.

Is Aeon a ChatGPT alternative?

For chatting, no. Keep ChatGPT for that. But if you have a routine you run through ChatGPT by hand every morning (paste, prompt, copy, repeat), Aeon turns that into a scheduled skill that reports back to you.

Can Aeon use OpenAI models?

Yes, via the Codex harness. Aeon runs skills through ten agent CLIs: Claude, Grok, Codex, Pi, Vibe, Kimi, fx, Cursor, Hermes, and GLM. Swap the harness in config and the skills don't change.

What does a skill look like?

One Markdown file: a small frontmatter block (name, schedule input, required keys) and a plain-language prompt. If you can write a decent ChatGPT prompt, you can write an Aeon skill.

Does "autonomous" mean no guardrails?

No. Read-only skills can't touch the repo, irreversible actions fail closed, secrets stay off the command line, and an optional authorization layer (Fleet Watcher) can gate every run if you want a human check.

Want to try it?

Point any coding agent at https://www.aeon.fun/skills/aeon.md and it sets Aeon up for you. Or fork aeonfun/aeon and run ./aeon.

Configure once, forget forever!