FOR OPEN-SOURCE AGENT AUTHORS
Your Agent repo, one command away from running.
Keep your Agent on GitHub. Frely gives it a page and a Deploy to Frely badge, and anyone can run it on their own computer inside a sandbox.
HOW IT WORKS
Three steps, and your README does the rest.
- 01
Add frely.yaml
A short file in your repository says how to start your Agent, and which extra sites it needs, if any.
- 02
List the repository
Paste the GitHub address on Frely. You get an Agent page under your name and a Deploy to Frely badge for your README.
- 03
Users run it locally
One command downloads your repository and starts it in a sandbox on their computer. Their API key never reaches your code.
How the sandbox protects their data
FAQ
Questions from Agent authors
What does my repository need?
A frely.yaml at the root with a local: section. Its run: command starts your Agent; setup: is optional and runs once after the download.
Does my code get to see the user’s API key?
No. Your Agent talks to a model address on the user’s computer. Frely’s command line tool adds the key itself and forwards only model requests, so your code never holds it.
What can my Agent do on the user’s computer?
It can change its own folder and the temporary folder, and reach Frely’s API. Any other site you list in frely.yaml needs the user’s approval first, once.
Is the code copied somewhere?
No. Frely keeps only the repository address, name, licence and the commit you listed. Users download from GitHub, and each run uses that commit until you refresh the listing.
Which computers can run it?
Where the Frely sandbox is available. If a computer cannot run it, the command stops instead of running your Agent without protection.
How the sandbox protects your dataALREADY ON GITHUB?
Add frely.yaml. Get a badge for your README.
Your repository stays yours. Frely lists it and gives your users a safe way to run it.
List your Agent