All-Agents documentation
What All-Agents does, how these docs are organised, and how to read them as a person or as an agent.
All-Agents turns a process you already run by hand into a workflow that runs itself. You describe the work in an interview, the platform watches a few real runs, and once the shape is stable it writes the repeatable parts down as code. The steps that need judgement stay with an agent. The steps that need a human stay with a human.
These docs describe what the platform does today. Nothing here is a roadmap. If a capability is documented, it is shipped and you can use it.
How the docs are organised
Workflows
Creating a workflow in the interview, teaching it through real runs, and what changes when it gets codified.
Triggers and schedules
What starts a run: a schedule, an event from another system, or you.
Runs and observability
What happened, in what order, what it cost, and how to replay it.
Approvals and gates
Where a human stays in the loop, and how a paused run comes back.
Integrations and code
Connecting the systems a workflow touches, and the code it is allowed to run.
Administration
Notifications, spend limits, and model selection.
The changelog records what shipped and when. The FAQ answers the questions that come up before the first workflow.
Reading these docs as an agent
Every page here has a plain-markdown twin, so an agent never has to parse HTML:
- Add
.mdto any docs URL./docs/approvals/permanent-gatesalso answers at/docs/approvals/permanent-gates.md. - Or ask for it. Send
Accept: text/markdownto the normal URL and you get the markdown back at the same address. - Start from the index.
/docs/llms.txtlists every page with its description and its.mdaddress, newest changelog entries first.
The twins are the same content as the page, with the interactive parts flattened
into text. They carry a noindex header, because the HTML page is the canonical
one.
When something is not here
These docs cover shipped behaviour only, so a gap means one of two things: the platform does not do it yet, or we have not written it up. Both are worth telling us about. Mail hey@all-agents.io and say what you were looking for.