Docs

Run Argus. Read the loop. Change the model.

Argus is one Worker with one durable Workflow. It gathers sources, follows weak spots, and returns a report with links.

01 · Local

Run it on your machine.

This builds the Worker and runs it locally. Workers AI still calls the real binding in your Cloudflare account.

02 · Deploy

Ship it to your account.

Use the Deploy to Cloudflare button in the README, or run the same steps yourself.

03 · The loop

Weak evidence creates the next search.

This is the core idea. Find thin spots, search again on purpose, then write the report from the board.

04 · Model

Set one Workers AI model.

ARGUS_MODEL controls the model used for source analysis and synthesis. The default is @cf/moonshotai/kimi-k2.6.

"vars": {
  "ARGUS_MODEL": "@cf/moonshotai/kimi-k2.6"
}