What’s New & Improvements
BYOK + New Plan Structure
Kody can now run using your own API keys, compatible with GPT-5, Claude 4.5, Gemini 2.5, Anthropic, or any other model you want.
You choose the provider, control the costs, and decide which model makes the most sense for your team.
Kody still handles the entire review flow, but now you manage the tokens and usage.
This gives you more flexibility, predictability, and transparency, along with the freedom to test different models and adjust Kody to what works best in your context.
Token Monitoring and Limits [Closed Beta]
Alongside BYOK, we added a complete monitoring dashboard to track token usage in real time.
The idea is simple: bring clarity and prevent surprises on the provider’s bill.
You can view usage by period, PR, or developer, set token limits, and get alerts when you’re close to the limit.
If you want to learn more about how to use BYOK, everything’s explained in the documentation.
New Plan and Pricing Structure
With BYOK’s arrival, we’ve redesigned Kodus plans to make everything simpler and more flexible.
- Community (Free): Previously, free access to Kody was only available for self-hosted setups. Now it’s also available in the cloud, with no limits on reviews, users, or issues. Just connect your API keys and start using it.
- Teams: Full-featured plan with analytics, autofix, support, and two usage options: BYOK (10 USD/dev/month) or Managed (30 USD/dev/month).
- Enterprise: Built for large teams, with RBAC, SSO/SAML, and dedicated SLAs.
All plans work with your own API keys (BYOK).
More details about the plans are available in the documentation.
See how each model performs in the Code Review LLMs Benchmark
Context Engine (Repo Files)
Kody can now use real files from your repository as context for rules and custom prompts.
That means when you create a rule or edit a prompt (for category, severity, suggestion, or PR summary), you can reference files from your own codebase to help Kody better understand your project’s structure and patterns.
To do this, it’s simple:
- Use @file:path/to/file.ts to reference a file from the same repository.
- Or @repo:org/project to reference files from another repository.
This new context mechanism makes reviews much more accurate, because Kody can now use your team’s actual code as a base for analysis.
You can learn more here: external rule references and external prompt references.
Customize Suggestion Messages
You can now customize Kody’s suggestion messages to match your team’s tone — from the content to the structure.
You can adjust the tone, style, and even the Markdown format of the messages Kody uses when suggesting improvements or fixes. This way, the suggestions align better with how your team communicates — they can be more direct, more detailed, or super short, depending on what makes sense for you.
Category Prompt Configuration
Before, the prompts defining the focus of each analysis type (like Bug, Security, or Performance) were fixed. With this update, your team can edit these sections however they like, defining what Kody should prioritize and how to classify issue severity.
This change gives you much more control to adapt Kody’s reviews to your code context, stack, and internal team standards.
Custom Plugins
You can now connect external MCP servers directly to your workspace, expanding what Kody can do during reviews.
Previously, only plugins from the official catalog (like Jira, Notion, or Linear) were supported. Now each team can add their own MCPs, connecting internal tools, private APIs, or custom workflows.
The plugin appears automatically in the workspace and follows the same security standards as official MCPs — Kody only accesses the endpoints you configure.
This update gives you more freedom to integrate Kody in a way that fits your team’s workflow.
Kody Issues + Tools
Kody Issues are now much more flexible and come with new control and integration options.
You can now create issues manually through the interface or via MCP, and also set filters to decide when automatic issues should be created.
Each issue now displays two new pieces of information:
- Owner: who opened the PR where the issue was found
- Reporter: who created the issue — it can be Kody (if automatic) or a user (if manual)
Kody can also create, edit, and delete issues via MCP, enabling automations and integrations with external tools.
Choose What to Monitor in the Cockpit
You can now control what appears in the Cockpit by selecting which metrics you want to see in the summary and detailed charts.
It’s possible to turn indicators like Deploy Frequency, PR Cycle Time, Bug Ratio, PR Size, and others on or off, keeping the dashboard focused only on what really matters for your team.
This way, each team decides what to track and has full control over the data that makes up their dashboard.
Improved Git License Management
The license screen is now much simpler and more practical.
You can now select users directly from your Git organization, without relying only on those who have already opened PRs. If the full org listing isn’t available, we use developers who have recently made PRs as a fallback, ensuring that all eligible users appear for selection.
We also added an automatic mode that fills licenses with the first developers identified through PRs, without the need to select everything manually.
This update makes license management easier and gives you more control over who receives Kody’s reviews.
Folder Rule Inheritance
We updated the rule inheritance system to reflect exactly how it works:
Global → Repository → Folder.
You can now also ignore inherited rules that don’t make sense at a given level. For example, a global rule can be disregarded in a repository, or a repository rule can be ignored in a specific folder.
This makes behavior more predictable and gives you more control over which rules actually apply in each context.
Bug Fixes
- ESC now works properly on the Issues screen
- Cross-file suggestions now display the correct severity
- File-level conversion is working normally again
