Overview
The Coding Agent is an AI-powered development assistant that helps developers build applications on the Condense platform. It can generate source code, create project structures, manage files and folders, generate automated test cases, and integrate directly with GitHub or GitLab repositories. The Coding Agent combines planning, code generation, and testing capabilities to accelerate application development while following the project structure required by Condense. With v2.6.0, the Coding Agent comes ready to use inside the built-in VS Code IDE, shows how much AI model capacity each environment uses, and includes a QA Agent that writes and checks tests for the code it generates. See What’s New in v2.6.0 below.Available Agents
The Coding Agent uses multiple specialized AI agents to complete development tasks.What’s New in v2.6.0
Three additions make the Coding Agent easier to reach, easier to monitor and easier to trust.Coding Agent ready in the VS Code IDE
Visual Studio Code (VS Code) is a widely used code editor. When you create an application in Condense, a web-based VS Code IDE starts automatically, and the Coding Agent is already configured inside it.- Nothing to install. The IDE opens in your browser.
- No separate sign-in. The Coding Agent uses your existing Condense session.
- The same Coding Agent. You can do everything described on this page from the IDE: generate code, manage project files, generate and validate tests, and work with GitHub or GitLab.
Token usage tracking
An AI model reads and writes text in small pieces called tokens. Every request you send to the Coding Agent, and every answer it writes, is counted in tokens, and tokens are the unit AI usage is measured in. The Coding Agent now tracks token usage. Tokens you send are counted as input tokens, and tokens the agent writes are counted as output tokens. In an environment, open Settings → Agents Usage to see its token usage, so you know how much AI model capacity your coding work consumes. Access to this page follows your role.QA Agent for generated code
A test case is a small automatic check that confirms code does what it should. After the Coding Agent writes code, the QA Agent generates test cases for that code and validates them. Writing the code and checking it now happen in the same workflow. You get the generated code together with its test results, so you can judge its quality before you deploy, instead of writing the checks by hand afterwards.Capabilities
The Coding Agent can assist with:- Generating Condense Applications
- Creating Custom Transforms
- Creating Connectors
- Generating project structures
- Managing project files
- Creating folders
- Updating existing source code
- Generating and validating automated test cases
- Integrating with GitHub repositories
- Integrating with GitLab repositories
Local Development Tools
The Coding Agent provides several tools for managing project files and folders.Development Workflow
A typical Coding Agent workflow consists of the following stages:- Analyze the development request.
- Create an implementation plan using the Sequential Thinking Agent.
- Generate the required source code.
- Create the necessary project files and folders.
- Generate test cases with the QA Agent and validate them against the generated code.
- Optionally publish the project to GitHub or GitLab.
Typical Use Cases
The Coding Agent can be used to:- Create new Condense Applications.
- Develop custom transforms.
- Build custom connectors.
- Scaffold new projects.
- Refactor existing code.
- Generate automated test cases.
- Maintain project structures.
- Publish code to Git repositories.
Best Practices
- Review generated code before deployment.
- Validate generated test cases.
- Store projects in version control.
- Follow your organization’s development standards.
- Use GitHub or GitLab integration for collaborative development.
Related Documentation
- GitHub Integration
- GitLab Integration
- Applications
- Custom Transforms
- Connectors