Create and Edit Agents

Build custom AI workflows with prompts, rules, and tools through the Continue Hub interface

Creating Custom Agents

You can only create and edit Agents through Mission Control in the Continue Hub web interface. This ensures proper validation, versioning, and team collaboration features.

Try an Agent First: Your 60-Second Challenge

Before creating your own agent, let’s see one in action! The fastest way to experience the power of Continue agents is with our demo repository.

See an Agent Create a Pull Request in Under 60 Seconds

1. 🍴 Fork Our Demo Repository

   Get a safe sandbox to experiment with:

   Option 1: GitHub Web Interface Visit github.com/continuedev/demo-project and click Fork

   Option 2: GitHub CLI

   gh repo fork continuedev/demo-project
   

2. 🎯 Access Mission Control

   Go to hub.continue.dev/agents and:

   • Connect GitHub and authorize Continue when prompted
   • This gives agents access to create PRs in your repositories
3. 🚀 Run Your First Agent

   Target your forked demo repo and try one of these commands:

   Add New Feature

   Create a new function that calculates fibonacci numbers.
   

4. ✨ Watch the Magic

   Result: A fully-formed Pull Request opens in your demo repo with:

   • Complete implementation or fix
   • Proper commit messages
   • Detailed PR description
   • Ready for review and merge

Why Use the Demo Repo?

• Safe testing environment - No risk to your production code
• Pre-configured issues - Common bugs and features to practice with
• Immediate results - See agents in action without setup complexity
• Learn by example - Study the generated code and PR descriptions

Once you see how agents work with the demo repo, you’ll understand exactly how to create and customize your own for real projects!

---

Creating an Agent

1. 🧭 Navigate to Create Agent

   From the Continue Hub top navigation bar, select ”+” → “New Agent”.

   This opens the Create an Agent (beta) form with all required fields.

2. 🧩 Fill Out the Agent Form

   Configure your agent with the following fields:

   Field	What to Enter	Example
   Name	Display name shown in the Hub	GitHub PR Agent
   Prompt	First instruction the agent receives	Open a GitHub PR to fix the specified issue.
   Description	What the agent does	Creates a pull request and includes AI-generated summaries.
   Tools (MCPs)	Select built-in or custom MCPs	GitHub, PostHog, Supabase
   Rules	Add any organizational rules	continuedev/gh-pr-commit-workflow, continuedev/summarization
   Model	Choose a default LLM	Claude Sonnet 4.5
   Owner + Slug	Determines namespace	my-org/github-pr-agent
   Visibility	Access level	Public, Organization, or Private

   Tip: Start with a simple prompt and add complexity through rules and tools. The prompt should be a clear, single instruction that defines the agent’s primary goal.

3. 💡 Preview the Configuration

   As you fill in the fields, the right-hand panel shows a live preview of your agent’s YAML configuration:

   name: Open PR with Fix
   description: Open a GitHub PR to fix the specified issue
   tools: built_in, anthropic/github-mcp
   rules:
     - continuedev/gh-pr-commit-workflow
     - continuedev/summarization
   model: Claude Sonnet 4.5
   visibility: public
   

   This preview helps you verify the configuration before creating the agent.

4. ✅ Create and Test

   Click “Create Agent” to save and publish it.

   Your agent is immediately available to run in:

   • Mission Control web interface
   • TUI mode: cn --agent your-org/your-agent-name
   • Headless mode: cn --agent -p your-org/your-agent-name "prompt" --auto

Editing an Agent

You can edit any agent you own or that belongs to your organization.

1. 🧭 Access the Agent

   From the Agents page or Mission Control view, click your agent’s name, then select “Edit Agent”.

2. 🔧 Update Configuration

   You can modify any of these components:

   Prompt

   Refine task behavior

   # Before
   Fix the bug in the authentication system
   
   # After  
   Fix the authentication bug by:
   1. Identifying the root cause
   2. Implementing a secure solution
   3. Adding appropriate tests
   4. Opening a PR with detailed explanation
   

3. 💾 Save Changes

   Click “Update Agent” when finished.

   The updated version is instantly available to your team with automatic versioning for change tracking.

Test your agent thoroughly to ensure they interact as expected.

Example Agent Configurations

Here are proven agent configurations you can create or use as inspiration:

Security Scanner Agent

Snyk Continuous AI Agent - Comprehensive security scanning with Snyk MCP integration.

Automates dependency analysis, vulnerability scanning, and creates remediation PRs with AI-powered fix suggestions.

Performance Monitor

Netlify Continuous AI Agent - Performance optimization with A/B testing and monitoring.

Tracks Core Web Vitals, identifies regressions, and provides optimization recommendations.

Error Monitoring Agent

Sentry Continuous AI Agent - Automated error analysis and issue creation.

Monitors production errors, provides root cause analysis, and creates actionable GitHub issues.

GitHub Management

GitHub Manager AI Agent - Comprehensive GitHub workflow automation.

Handles issue triage, PR reviews, and release note generation with natural language prompts.

Database Management

Supabase Continuous AI Agent - Database security and management workflows.

Audits Row Level Security, identifies vulnerabilities, and generates fixes automatically.

Troubleshooting

Common issues and solutions:

Problem	Solution
Agent doesn’t complete tasks	Simplify the prompt, add more specific instructions, verify tool permissions
Tools aren’t working	Check MCP configuration, verify API keys and permissions in Hub settings
Inconsistent behavior	Add rules to enforce consistent patterns, test with various input scenarios
Performance issues	Consider model choice, simplify complex multi-step workflows, optimize tool usage