All articles
Getting Started
What is CICDoo Getting started: the onboarding wizard The Dashboard and Infrastructure MapConcepts
Production, staging and development stages Domains, subdomains and SSL Odoo versions and editions (Community vs Enterprise) Workspaces and team collaborationServers
Adding a server Server domains and DNS Using a load balancer Server actions: deploy, logs and charts Deleting a serverProjects & Git
Creating a project Connecting GitHub Project GitHub settings and webhook Connecting GitLab Branches and instances Tagging and filtering projects High availability with app nodes Project environments, extensions and deletionInstances & Console
Creating an instance Restarting, stopping and removing an instance Merging branches between stages Web IDE and remote access SSH access to instances Instance settings explained Deployments and the job queueAgent Tasks
What agent tasks are Setting up agent tasks Running a task and reviewing the result Agent time and top-ups Agent insights and task filtersMonitoring & Backups
Monitoring your instance Backups and restore S3 storage, uploads and bulk backup actions Alert notifications (email and SMS)Workspaces & Permissions
Inviting workspace members Member permissions and resource access A member cannot see or use a resource Audit log and server activityAccount & Security
Two-factor authentication (2FA) Profile and integrations settings Passwords and account recoveryBilling & Plans
Plans and pricing Upgrading and managing your subscription If a project subscription goes unpaid The channel partner program Onboarding a client as a partnerSupport & Tickets
Opening a support ticket Getting help from the CICDoo teamTroubleshooting
Troubleshooting: cannot connect a server Troubleshooting: a deployment failed Troubleshooting: git push fails with 403 after moving a repository Troubleshooting: my instance is down or slow Troubleshooting: domain or SSL problemsAgent Tasks
What agent tasks are
Describe a change in plain language and a coding agent plans it, waits for your approval, then implements it inside the instance you picked. What the agent can and cannot do, and the rules that protect production.
What an agent task is
An agent task is a piece of work you describe in plain language. A coding agent picks it up inside one of your own instances, works out how to do it, and waits for you to approve that plan before it writes any code. Once you approve, it implements the change in that instance and pushes the result to the instance's branch.
You are not handing work to a service somewhere else. The agent runs in your instance's container, against your code, on the branch that instance is deployed from.
The three steps
Every task moves through the same three steps, and you are in the middle of them.
- You describe the work. You give a title and a description of what should change. Plain language is enough. You pick which instance the agent works in, and you can attach files for context.
- The agent plans. It reads the codebase in that instance and writes back a plan: what it intends to change and how. Nothing is written yet. If the ask is unclear or it cannot do the work safely, it says so instead of guessing, and the task waits for you.
- You approve, then it implements. Once you click "Approve & Execute", the agent makes the change, commits it and pushes to the instance's branch. That push redeploys the instance, so you can look at the result running.
If the result is not what you wanted, you send it back with "Start Another Round". The agent builds on the code it already pushed rather than starting over, and you approve the new plan before anything runs again.
Where to find agent tasks
- For one project: open the project console and click "Agent Tasks" in the sidebar. This page lists the project's tasks and has the "New Task" button.
- For one instance: in the instance console, the "Agent Tasks" tab lists tasks for that instance only. The tab appears once agent tasks are enabled on the project.
- Across all projects: the "Agents" item in the main menu opens a page with charts and a task list covering every project in the workspace. See Agent insights and task filters.
You see these only if your workspace role includes the "Agent tasks" permission. Approving a plan needs the separate "Approve plans" permission.
Where an agent can work
Agent tasks run on the instances of a project you have enabled them for. Staging and development instances are the normal place for this work: enabling agent tasks starts a poller inside each of them, ready to pick tasks up.
Production is allowed, but deliberately awkward. A task aimed at a production instance has to be confirmed by typing a word when you create it, and again when you approve its plan. Its plan always waits for a person to read it: auto-execute, which lets a plan run without review, is refused outright on production. There is no combination of settings that lets an agent change production while nobody is watching.
What the agent produces is yours to check
The agent writes generated code. It can be wrong, or incomplete, in ways that are not obvious from reading the plan. Treat its output the way you would treat a pull request from someone who has never seen your business:
- Take a current backup before approving anything on production.
- Have someone who knows the codebase read the plan, not just the summary.
- Test the result on the instance before you carry it forward.
CICDoo provides the agent as a tool. The changes that get made are the ones you approved, and the outcome of those changes is yours. This is set out in section 11 of the Terms of Service.
What it is not
- Not a chatbot. It does not answer questions about your data; it changes code in an instance.
- Not the AI log analysis feature. That is a separate thing, for reading deploy and server logs.
- Not a shared service. The agent runs on your own provider account, using a credential you supply, so runs count against your own subscription or quota with that provider.
Related: Setting up agent tasks, Running a task and reviewing the result, Agent time and top-ups, Agent insights and task filters.
If you get stuck, open a support ticket from the Tickets page or the chat widget.
Still stuck? Open a ticket from the app or talk to an engineer.