Here you’ll find the core functionalities that make Blar feel like an extra engineer on your team. From bug detection to enforcing your own design patterns, every feature is built to help you write better code, faster.
We're constantly evolving shipping new capabilities all the time and your feedback plays a key role. If there’s something you need, let us know. Some of our best features started as user requests 😉
At Blar, we have five different types of agents, each focused on a specific aspect of code quality: bugs, security, performance, design consistency, and maintainability. You can fully customize which agents to activate in each repository, and also control the level of comments they leave, whether it's just info, warnings, or errors.
Detects potential bugs, inconsistencies, and edge cases before they hit production. This agent dives deep into your pull requests to spot logic flaws, risky conditions, and behaviors that may break under certain inputs or scenarios. It acts as your first line of defense against those tricky issues that are hard to catch with tests alone.
Scans your code for known vulnerabilities, insecure practices, and potential attack vectors. Whether it's improper input sanitization, weak cryptographic usage, or risky dependency calls, this agent helps you ship more secure code without needing a security expert on every review.
Focuses on performance bottlenecks, especially around database usage. It flags inefficient queries, unindexed filters, or N+1 problems that could slow down your app in production. Ideal for maintaining scalability as your user base grows.
Checks that your team’s code adheres to your documented design patterns. It uses your internal Wiki as a reference and alerts you when someone deviates from agreed standards. Great for keeping code consistency and reducing ramp-up time for new developers.
Sniffs out design anti-patterns that hurt maintainability—things like classes with too many methods, functions with too many parameters, or duplicate logic. It encourages cleaner, more modular code and helps prevent long-term technical debt from creeping in.
Blar doesn't just review your code—it continuously learns alongside you and your team. Every interaction counts, from specific comments you leave about your code to detailed feedback on suggested improvements (yes, included thumbs down and thumbs up 😅). Blar even picks up on edge cases and specific business rules you share, ensuring future recommendations become increasingly accurate and perfectly tailored to your coding style and specific requirements.
Blar has several ways to learn:
Learns from your feedback:
When you give a comment a 👍 or 👎, Blar records that to improve future reviews.
Learns from your rules (Wiki):
You can define your technical standards in a Wiki. Blar uses it as a guide during every review.
Learns from your conversations:
If you chat with Blar inside a PR (e.g., asking it to ignore specific rules), it will remember that behavior for future reviews in the same context (function, repo, or team).
Chat directly with Blar’s intelligent agents, designed to understand your codebase and help you move faster. Whether you’re trying to debug a persistent issue, define custom rules, or understand why a fix was suggested, our agents deliver clear answers and real-time solutions. Think of it as having a bug-squashing senior engineer on call 24/7, but faster.
You can interact with Blar in several ways:
Command comments:
blar --review: reviews new commits.blar --review --force: reviews the entire PR.blar --fix: suggests an automatic fix.Contextual chat:
Use it like a human reviewer — ask for examples, explanations, or rule exceptions.
Reactions (feedback):
Use 👍 or 👎 on Blar's comments to help it learn and improve.