> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cloudthinker.io/llms.txt
> Use this file to discover all available pages before exploring further.
# Autonomous Operations
> Learn what CloudThinker runs on its own — keepers, scheduled tasks, Pulse triage — and how Manual and Auto modes keep you in control
CloudThinker agents don't just answer prompts — they monitor, analyze, and act on your infrastructure continuously. This page explains what runs autonomously and how the two autonomy modes keep every action under your control.
## Why autonomous operations
* **Problems get caught early.** Keepers scan connected accounts continuously, so drift surfaces in minutes instead of at month-end review.
* **Routine work runs itself.** Scheduled tasks handle daily cost checks, weekly audits, and monthly reports without anyone remembering them.
* **Noise becomes signal.** Pulse suppresses duplicate monitoring events and ranks what's left, so on-call engineers see clusters worth acting on.
* **You set the risk boundary.** Every write action follows your Manual or Auto setting — agents never gain permissions you didn't grant.
## The autonomy model
Every agent write action runs in exactly one of two modes:
| Mode | What happens |
| ---------- | ---------------------------------------------------------------------------------- |
| **Manual** | The agent proposes the action and waits for a person to approve it before running. |
| **Auto** | The agent runs the action on its own and reports the result. |
Read operations never require approval. You decide which writes wait for a person:
* [Approval](/guide/approval) sets per-tool permissions and approvers for your workspace.
* [Auto Mode](/guide/auto-mode) adds an intent classifier that auto-approves writes matching your request and pauses anything outside it.
Start new keepers and scheduled tasks in Manual. Switch individual actions to Auto once you've reviewed a few runs and trust the pattern.
## What runs autonomously
Three surfaces do work without a prompt from you. Each is documented on its own page — configure them there.
| Surface | What it does on its own | Learn more |
| -------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | --------------------------------------------- |
| **CloudKeepers** (shown as **Keepers** in the app) | Monitor connected accounts and clusters for cost, security, and performance drift, then raise findings and tracked recommendations | [Keepers](/guide/infrastructure/cloudkeepers) |
| **Scheduled tasks** | Run an agent prompt on a daily, weekly, or monthly schedule — cost analyses, security audits, executive reports | [Tasks & Scheduling](/guide/automation/tasks) |
| **Pulse** | Ingests events from sources like CloudTrail, GuardDuty, and Datadog, suppresses noise, and surfaces ranked clusters that can escalate to [incidents](/guide/incident/overview) | [How Pulse Works](/guide/pulse/overview) |
Whichever surface triggers the work, the same rule applies: each write action either waits for approval or runs and reports, per your autonomy settings.
## Stay informed
Every autonomous run is logged in your workspace, and [notifications](/guide/notifications) deliver findings, errors, and approval requests to Slack, Teams, or email. Review activity regularly and tighten or loosen autonomy as patterns emerge.
## Related
Configure per-tool permissions and approvers for agent actions
Let a classifier auto-approve writes that match your intent
Schedule agent prompts to run daily, weekly, or monthly
Enable autonomous monitors for cost, security, and performance drift