Files
continue/docs/mission-control/tasks.mdx
BekahHW 0e28c6f1e5 docs: cloud agents improvement (#9443)
* Change title and description

* Add cloud agents definition

* Add comparison table for local v. cloud

* Add more targeted description

* Optimize card language

* Update description to include cloud agents

* Optimize mission control page for cloud agents

* More CA optimization

* Add early distinction

* Optimize for CA and remove references to UI elements no longer there.

* Clean up into to agents and optimize

* Optimize

* Shorten title and fix link

* Update Integrations with cloud agents

* Add top asked for cloud agent feature to GH docs

* Add FAQs to GitHub integration page

* Optimize metrics page for cloud agents

* Replace old new agent link

* Update docs/mission-control/metrics.mdx

Co-authored-by: cubic-dev-ai[bot] <191113872+cubic-dev-ai[bot]@users.noreply.github.com>

* Update docs/agents/intro.mdx

Co-authored-by: cubic-dev-ai[bot] <191113872+cubic-dev-ai[bot]@users.noreply.github.com>

* Update docs/agents/intro.mdx

Co-authored-by: cubic-dev-ai[bot] <191113872+cubic-dev-ai[bot]@users.noreply.github.com>

---------

Co-authored-by: cubic-dev-ai[bot] <191113872+cubic-dev-ai[bot]@users.noreply.github.com>
2026-01-07 15:54:53 -08:00

138 lines
2.8 KiB
Plaintext

---
title: "Tasks"
description: "A Task is a unit of work shared between you and an agent. You trigger the task, the agent executes it, then you review or approve the results. Tasks can run **on demand**, **on triggers**, or **on schedules**."
---
<Info>
**What is a Task?**
A Task is a unit of work that combines:
- Your instructions
- An agent's capabilities (model, rules, tools, prompt)
- Your codebase (repository and branch)
Tasks can be triggered manually, by events, or on a schedule, and can be assigned to anyone on your team for review.
</Info>
## Creating and Running a Task
<Steps>
<Step title="Choose an Agent">
Select the **Agent** you want to use for this task.
<Tip>
Don't have an Agent yet? Create one in **[Mission Control → Agents](https://hub.continue.dev/agents/new)** before setting up your first task.
</Tip>
</Step>
<Step title="Select Your Repository">
Pick the **repository** and **branch** the Agent branch off of.
</Step>
<Step title="Describe the Task">
Add your user input — clear instructions for what you want the agent to accomplish.
**Example prompts:**
```text
Refactor unused React components and clean up imports
```
```text
Add error handling to all API endpoints in the /api directory
```
```text
Update README with installation instructions for the new CLI tool
```
</Step>
<Step title="Run the Task">
Click the Submit button.
The Agent will:
- Analyze your repository and branch
- Execute the task based on your input
- Generate changes based on your task request
- Create a detailed session log
<Info>
Task execution time varies based on complexity. You can monitor progress in your inbox.
</Info>
</Step>
<Step title="Review Results">
Each Task run creates a Session with complete traceability.
<AccordionGroup>
<Accordion title="Summary">
A high-level explanation of:
- What the Agent did
- Why decisions were made
- Key changes implemented
</Accordion>
<Accordion title="Diff">
The exact code changes:
- Files modified
- Lines added/removed
- Side-by-side comparison
</Accordion>
<Accordion title="Logs">
Full execution trace:
- Tool calls made
- Agent reasoning
- Any errors or warnings
</Accordion>
</AccordionGroup>
<Tip>
Review the diff carefully before accepting changes. You can:
- Accept all changes
- Request modifications
- Reject and retry with an updated input
</Tip>
</Step>
</Steps>
## Next Steps
<Card title="Create a Workflow" icon="workflow" href="/mission-control/workflows">
Automate tasks with event-based triggers
</Card>