In December, we launched Conductor, an extension for the Gemini CLI designed to carry context-driven growth to your terminal. By shifting undertaking consciousness out of ephemeral chat logs and into persistent, version-controlled markdown information, Conductor has helped builders worldwide plan earlier than they construct.
Right now, we’re releasing a brand new function to assist make AI-assisted engineering safer and extra predictable. Our new Automated Evaluate function permits Conductor to transcend planning and execution into validation, producing post-implementation experiences on code high quality and compliance to the rules you’ve outlined.
Shut the loop with Automated Evaluations
Our new Automated Evaluate function introduces a rigorous “confirm” step to the event lifecycle. As soon as your coding agent completes its duties, Conductor can now generate a complete post-implementation report.
Key capabilities:
- Code evaluate: Conductor acts as a peer reviewer, performing deep static and logic evaluation on newly generated information. Past simply syntax, it proactively flags advanced points equivalent to race circumstances in asynchronous blocks, potential null pointer dangers, and logic errors that would result in runtime exceptions
- Plan compliance: The system routinely checks the brand new code towards your plan.md and spec.md. It ensures each section of your roadmap was addressed and that no core necessities had been omitted throughout the coding course of.
- Guideline enforcement: To keep up long-term code well being, Conductor verifies that every one new contributions adhere strictly to your undertaking’s particular fashion guides in addition to any customized guideline information generated throughout the planning section.
- Check-suite validation: As a substitute of counting on guide execution, Conductor integrates your total take a look at suite straight into the evaluate workflow. It runs all related unit and integration assessments, then incorporates the outcomes and protection knowledge into the ultimate report to offer a unified view of whether or not the brand new code truly features as supposed inside your current ecosystem.
- Fundamental safety evaluate: Security is built-in into the core of each evaluate, with the system scanning for vital vulnerabilities earlier than code is ever merged. It routinely flags high-risk points equivalent to hardcoded API keys, potential PII (Personally Identifiable Info) leaks, or unsafe enter dealing with that would expose the applying to injection assaults.
Actionable intelligence
The output of an Automated Evaluate is designed for motion. Findings are categorized by severity (Excessive, Medium, and Low), offering builders with clear directions on the place to iterate. You’ll obtain the precise file path and may begin a monitor inside Conductor to repair the problems.
This degree of element ensures that “agentic” growth doesn’t suggest “unsupervised” growth. As a substitute, it creates a workflow the place the AI offers the labor and the developer offers the high-level architectural oversight, backed by automated verification.
Get began
Conductor is evolving quickly, and these options signify our dedication to creating AI growth secure, predictable, and architecturally sound.
Set up the extension and begin utilizing Conductor at the moment right here or by utilizing the next command:
gemini extensions set up https://github.com/gemini-cli-extensions/conductor
