# All-Agents documentation

> All-Agents turns a process you run by hand into a workflow that runs itself: an interview captures the work, real runs teach it, and the repeatable parts get written down as code. These docs describe shipped behaviour only, so a capability that is absent from them is one you should not assume exists.

## Markdown access

Every page below is linked as its plain-markdown twin. The same content is also served from the HTML URL when the request carries `Accept: text/markdown`, so `https://www.all-agents.io/docs/workflows` and `https://www.all-agents.io/docs/workflows.md` return the same markdown.

Something missing or wrong? Mail hey@all-agents.io and say what you were looking for.

## Docs

- [All-Agents documentation](https://www.all-agents.io/docs.md): What All-Agents does, how these docs are organised, and how to read them as a person or as an agent.
- [Administration](https://www.all-agents.io/docs/administration.md): Account-level controls: what All-Agents tells you about, what it is allowed to spend, and which model it runs on.
- [Notifications](https://www.all-agents.io/docs/administration/notifications.md): The moments All-Agents emails you about, the ones it deliberately stays quiet on, and where to read back everything it has sent.
- [Spend limits](https://www.all-agents.io/docs/administration/spend-limits.md): Cap what your workflows may spend in a month, at the workspace, account or single workflow level, and see where the month went.
- [Approvals and gates](https://www.all-agents.io/docs/approvals.md): The two ways a person stays in the loop on a run, what a paused run looks like, and how it comes back.
- [Approval requests](https://www.all-agents.io/docs/approvals/approval-requests.md): How an agent stops mid-run to ask you before it does something consequential, and what happens when you answer.
- [Permanent gates](https://www.all-agents.io/docs/approvals/permanent-gates.md): A checkpoint written into a workflow that stops every run for a person, even once the workflow is fully automatic.
- [Responding to an approval](https://www.all-agents.io/docs/approvals/responding-to-approvals.md): Where pending decisions land, what a card shows you, what you can change before approving, and what a rejection does.
- [Integrations and code](https://www.all-agents.io/docs/integrations.md): How a workflow reaches the systems you work in, and what the code it runs is allowed to touch.
- [Adding an integration](https://www.all-agents.io/docs/integrations/adding-an-integration.md): What happens when a workflow needs a system it cannot reach yet, and why nothing has to ship on our side first.
- [Connecting accounts](https://www.all-agents.io/docs/integrations/connecting-accounts.md): How a workflow gets into your CRM, your inbox or an internal portal, and what happens to the credential afterwards.
- [Models and your own key](https://www.all-agents.io/docs/integrations/model-selection-and-byok.md): Choosing the model a workflow runs on, and running everything of yours on your own OpenAI key.
- [What a workflow's code may do](https://www.all-agents.io/docs/integrations/tenant-code-and-capabilities.md): Every codified step declares what it can reach before it runs, and the console shows you that declaration step by step.
- [Runs and observability](https://www.all-agents.io/docs/runs.md): What a run records, where to find it, and how to read the whole story after it finished.
- [Run artifacts](https://www.all-agents.io/docs/runs/run-artifacts.md): The browser recordings and collected files a run leaves behind, where they appear, and how to get them.
- [Stopping a run](https://www.all-agents.io/docs/runs/run-control.md): The emergency brake on the run screen, what it stops, what it cannot stop, and how to stop a whole workflow at once.
- [The run log](https://www.all-agents.io/docs/runs/run-log.md): One append-only record per run, read two ways as the timeline and as the transcript.
- [Test runs](https://www.all-agents.io/docs/runs/test-runs.md): How a test fire differs from a real one, what your note and payload can steer, and what a test never counts toward.
- [Watching a run](https://www.all-agents.io/docs/runs/watching-a-run.md): The live timeline, the working indicator on the lists, and looking over the agent's shoulder in the browser.
- [Triggers and schedules](https://www.all-agents.io/docs/triggers.md): The three things that can start a run, how a trigger gets armed and disarmed, and what happens when the same event arrives twice.
- [Email triggers](https://www.all-agents.io/docs/triggers/email-triggers.md): How a workflow runs when a message arrives in your mailbox, which providers connect with an app password, what the run receives, and why Microsoft mail cannot do this yet.
- [Event triggers](https://www.all-agents.io/docs/triggers/event-triggers.md): How a workflow runs when something happens in another system, why there is no catalogue of ready-made integrations, and how a broken trigger gets repaired.
- [Running a workflow yourself](https://www.all-agents.io/docs/triggers/manual-runs.md): The Run now button, when it appears, what it refuses, and why a real run and a test run are counted differently.
- [Trigger payloads](https://www.all-agents.io/docs/triggers/payloads.md): What a run receives when it fires, which keys the platform reserves, and why a payload that would have to be guessed at is refused instead.
- [Schedules](https://www.all-agents.io/docs/triggers/schedules.md): How a recurring schedule is set from plain words, how timezones and daylight saving behave, and what happens when a schedule stops making sense.
- [Sample payloads and test runs](https://www.all-agents.io/docs/triggers/testing.md): How to fire an event-driven workflow against a realistic payload before anything real reaches it, and what a test run will refuse to invent.
- [Webhook URL triggers](https://www.all-agents.io/docs/triggers/webhook-url.md): A URL you paste into another service so its deliveries start your workflow, how a listen window captures a real payload before you publish, and the auth token that comes with it.
- [Workflows](https://www.all-agents.io/docs/workflows.md): How a workflow comes into existence in an interview, learns from real runs, and becomes deterministic code once you decide it is ready.
- [Codification](https://www.all-agents.io/docs/workflows/codification.md): The walk-through that turns accepted runs into deterministic steps, what you can pin along the way, and what changes once you confirm.
- [Creating a workflow](https://www.all-agents.io/docs/workflows/creating-a-workflow.md): Name the job, talk it through with the assistant, then publish. What the setup interview writes and what Publish actually does.
- [Editing and revising](https://www.all-agents.io/docs/workflows/editing-and-revising.md): Changing a workflow on purpose, adapt or reset on a codified one, repairing a failed run, and stepping back to learning.
- [Reference files](https://www.all-agents.io/docs/workflows/knowledge-files.md): Upload the price list, policy PDF or template a workflow works from, and see how those files reach a run.
- [Teaching it through real runs](https://www.all-agents.io/docs/workflows/teaching-and-learning-runs.md): How a published workflow learns: sequential agentic runs, accepting or rejecting each one, and the readiness signals that say it is ready to codify.
- [The workflow directory](https://www.all-agents.io/docs/workflows/workflow-directory.md): A workflow is one versioned directory holding its written procedure, its steps and its reference files, under a single version number.

## Changelog

Newest first. A changelog entry is evidence that a capability shipped, not a description of how it behaves today.

- [Paste one URL, and the workflow listens before it runs](https://www.all-agents.io/docs/changelog/2026-08-27-paste-one-url-and-the-workflow-listens.md): A trigger can now be a webhook URL you paste into another service, with a listen window that captures a real delivery into the payload library before you publish.
- [Run a workflow when an email arrives](https://www.all-agents.io/docs/changelog/2026-08-27-run-a-workflow-when-an-email-arrives.md): Gmail, iCloud, GMX and WEB.DE mailboxes can start a workflow. You hand over the address and an app password, and the platform reads the mailbox itself.
- [Public docs you can read, search and ask questions of](https://www.all-agents.io/docs/changelog/2026-08-19-public-docs.md): Documentation lives at /docs, with search, an Ask AI box, and a plain markdown copy of every page for agents.
- [See what you have connected, and disconnect it](https://www.all-agents.io/docs/changelog/2026-08-19-see-and-disconnect-your-connections.md): Connections get a real list with a Disconnect button, any system can be named, and third-party browser logins now finish properly.
- [The interview stops guessing what the platform can do](https://www.all-agents.io/docs/changelog/2026-08-19-the-interview-stops-guessing.md): The agent that designs your workflow reads the published capability docs before it answers, and tells you when it could not.
- [Set a spending cap, and unattended work parks instead of running up a bill](https://www.all-agents.io/docs/changelog/2026-08-14-spend-limits.md): Monthly caps for your workspace, for you and for any single workflow, with a ledger behind them and a notification when one bites.
- [Files your agent downloads come back to you](https://www.all-agents.io/docs/changelog/2026-08-13-files-the-agent-downloads.md): A run can collect a file from a browser download and attach it to the run, with caps and an honest stub when a file is too large.
- [See everything a run did, in one place](https://www.all-agents.io/docs/changelog/2026-08-13-one-run-timeline.md): One timeline replaces the split feed and step trace, with full tool calls, durations, and the question each gate asked.
- [Schedules run in your timezone](https://www.all-agents.io/docs/changelog/2026-08-13-schedules-in-your-timezone.md): A schedule carries its own zone, shows its next run in local time, and keeps firing once a day across both daylight saving edges.
- [Stop a run, and correct a checkpoint instead of ending the run](https://www.all-agents.io/docs/changelog/2026-08-12-stop-a-run-and-correct-a-checkpoint.md): Cancel one run or every run of a workflow at once, and say no at a gate without throwing away the work behind it.
- [See what a run cost, in dollars](https://www.all-agents.io/docs/changelog/2026-08-10-what-a-run-cost.md): Each run carries an estimated cost beside its model, priced per model with the cached share of the input billed at the cached rate.
- [Fire a run yourself, for real or as a rehearsal](https://www.all-agents.io/docs/changelog/2026-08-09-fire-a-run-yourself.md): A play button runs the published workflow for real, and a test fire carries your own note and payload with your authority behind it.
- [Choose the model a workflow runs on, and pay with your own key](https://www.all-agents.io/docs/changelog/2026-08-09-pick-your-model-and-your-key.md): Set the model and reasoning effort per workflow, and route every agent of yours through your own OpenAI key.
- [Watch a recording of what the browser did](https://www.all-agents.io/docs/changelog/2026-08-09-watch-a-recording-of-the-browser.md): Every browser session a run opens is recorded, stored on our side, and playable from the run page days afterwards.
- [Read the exact conversation a run had](https://www.all-agents.io/docs/changelog/2026-08-08-read-the-run-transcript.md): The run page serves every turn, tool call and thinking step verbatim, instead of the clipped preview the live feed shows.
- [Everything a workflow is, in one version history](https://www.all-agents.io/docs/changelog/2026-08-07-one-version-history-per-workflow.md): The written procedure, the steps and your reference files move together under one counter, and a change asks whether to adapt or reset.
- [Your agents know what day it is](https://www.all-agents.io/docs/changelog/2026-08-07-your-agents-know-the-date.md): Every agent surface opens with the current UTC time, and a scheduled run is told which occurrence woke it.
- [One agent watches the run, and a failure opens a conversation to fix it](https://www.all-agents.io/docs/changelog/2026-08-06-one-agent-watches-the-run.md): Codified steps run with no model in the loop until something needs judgment, and a failed run opens a repair interview.
- [See what every automated step is allowed to reach](https://www.all-agents.io/docs/changelog/2026-08-05-see-what-every-step-can-reach.md): Each code step declares its network access and its connections, and the console prints the complete list beside your steps.
- [Your agent writes the integration for the system you actually use](https://www.all-agents.io/docs/changelog/2026-08-01-the-agent-writes-the-integration.md): The interview authors sandboxed code that subscribes to your service's events, so a trigger no longer waits on our backlog.
- [Workflows that fire when the thing happens, and a Pause button](https://www.all-agents.io/docs/changelog/2026-07-31-event-triggers-and-pause.md): Event triggers stay armed and repair themselves, Pause turns a workflow off without deleting it, and a pinned payload makes an event test real.
- [Watch what the agent is doing while it does it](https://www.all-agents.io/docs/changelog/2026-07-31-watch-the-agent-while-it-works.md): Tool calls now carry their arguments and results, the model's reasoning summaries are shown, and every list says what is busy.
- [Test run does exactly what you are looking at](https://www.all-agents.io/docs/changelog/2026-07-30-test-run-runs-what-you-are-looking-at.md): Fire the draft or the published version on purpose, get the refusal at the click, and see a trigger that failed before a run existed.
- [Close the tab and your run keeps going](https://www.all-agents.io/docs/changelog/2026-07-29-close-the-tab-and-the-run-keeps-going.md): Agent turns run detached from the browser request now, on a durable log you can reload straight back into.

## FAQ

- [Can a team share a workflow?](https://www.all-agents.io/docs/faq/can-a-team-share-a-workflow.md): No. A workflow has exactly one owner, who builds it, runs it and answers its approvals.
- [Can I stop a workflow running up a large bill?](https://www.all-agents.io/docs/faq/can-i-cap-what-it-spends.md): Yes. Set a monthly cap on one workflow or on everything you run, and unattended work parks instead of spending past it.
- [Can I see exactly what a run did?](https://www.all-agents.io/docs/faq/can-i-see-what-a-run-did.md): Yes, and an autonomous run is recorded the same way a supervised one is.
- [Can I stop a run that is already going?](https://www.all-agents.io/docs/faq/can-i-stop-a-run.md): Yes. You can cancel a live run from the screen you are on, and stop the whole workflow in the same action.
- [Can I test a workflow without touching real systems?](https://www.all-agents.io/docs/faq/can-i-test-without-touching-real-systems.md): A test run is a real run with real credentials and real side effects, and there is no separate staging environment.
- [Do I need to know how to code to build a workflow?](https://www.all-agents.io/docs/faq/do-i-need-to-code.md): No. You describe the process in an interview and correct the first real runs, and the platform writes the code.
- [Does it work with the tools we already use?](https://www.all-agents.io/docs/faq/does-it-work-with-our-tools.md): If the service has an API and you can give the workflow a credential for it, yes. There is no catalogue of supported apps.
- [Is there a library of ready made workflows to start from?](https://www.all-agents.io/docs/faq/is-there-a-library-of-ready-made-workflows.md): No. Every workflow starts from a name and the interview, and there are no prebuilt templates in the product.
- [What can start a workflow?](https://www.all-agents.io/docs/faq/what-can-start-a-workflow.md): A schedule, an event from a system you connected, or a person pressing run.
- [What does a single run cost?](https://www.all-agents.io/docs/faq/what-does-a-run-cost.md): It depends how much of the workflow is still agentic, and every run shows you its own number.
- [What happens when a workflow breaks?](https://www.all-agents.io/docs/faq/what-happens-when-a-workflow-breaks.md): The failure opens a repair interview, and you choose between a checked patch and sending the workflow back to learning.
- [What if the process is different depending on the case?](https://www.all-agents.io/docs/faq/what-if-the-process-differs-case-by-case.md): The steps run in the same order every time. Variation lives inside a step, and genuinely open judgement stays with an agent.
- [When does a workflow start running on its own?](https://www.all-agents.io/docs/faq/when-does-a-workflow-run-on-its-own.md): When you decide it is ready. No run count and no confidence score flips it for you.
- [Who can see the data our workflows handle?](https://www.all-agents.io/docs/faq/who-can-see-our-data.md): Each workflow runs in its own sandbox with only the credentials it declared, and it acts with its owner's access rather than a shared account.
- [Who gets asked when a run needs approval?](https://www.all-agents.io/docs/faq/who-gets-asked-to-approve.md): The workflow's owner, and only the owner. There are no approval chains, no escalation and no timeout.
- [Will it do something irreversible without asking me?](https://www.all-agents.io/docs/faq/will-it-act-without-asking-me.md): An agent step is instructed to stop and ask before anything consequential, and a gate you pinned always stops. Only the second one is a guarantee.