Codity Is Live on Product Hunt

Codity is live on Product Hunt today.

We have been building it for a while, quietly, with teams who let us sit inside their pull requests and tell us when we were wrong. Today it goes out properly. This post is the version of the launch that does not fit in a tagline: what Codity actually does, and the reasoning that put each piece in the product.

If you would rather skip the reasoning and go straight there: Codity on Product Hunt.

The problem we kept running into

Writing code stopped being the expensive part of software. Reading it did not.

A line gets written once and then read over and over, by people trying to change something near it. It gets read in review. It gets read at 2am by whoever is on call. It gets read a year later by someone who was not there for the decision and has to work out from the code alone whether that early return is load bearing.

AI assistants made the writing step close to free. They did not make any of the reading steps cheaper. So the ratio moved, and the pressure landed downstream, on review. We wrote about that in more detail in AI is writing more code than ever.

Most tooling answers that pressure by reading the diff faster. We think the diff is the wrong unit. A change is only safe or unsafe relative to the system around it, and the system is not in the diff.

So Codity indexes the repository first and reviews against that. Everything else in the product falls out of the same decision.

That is the whole argument, and it is the one we put up on Product Hunt today to be tested in public.

What shipped

Review that has read the rest of the codebase

Every pull request gets a plain-language summary on open, a merge readiness score, and findings ranked by what will actually break rather than by rule severity. Each finding carries a suggested fix you can commit from the review.

Two things we care about more than the finding count:

Requirement-aware review. If the PR is linked to a Jira ticket or a Confluence page, Codity reads the acceptance criteria and checks the change against them. "This is correct code" and "this is the change that was asked for" are different questions, and only one of them was ever being asked.

Confidence scoring and concise mode. A reviewer that comments on everything gets muted within a week. Findings are scored, low-confidence ones are held back, and concise mode collapses the rest into one consolidated comment instead of forty inline threads.

Security in the same pass

Secrets, injection and unsafe data handling, authentication gaps, vulnerable dependencies, and license compliance. All of it scanned on the change, and available as a whole-repository scan. Each finding is explained as an exploit path in plain language rather than a rule ID, because a rule ID is not something you can argue with at review time.

More on security scans.

A live map of the architecture

This is the piece we are least willing to give up. Codity builds a persistent, multi-repository map of your architecture: services, dependencies, the connections between repositories with a confidence percentage and the code that proves each one. You can correct it: describe a connection, reverse it, deny it. It keeps the history.

On top of the map is Intelligence Chat: ask a question in plain language across every repository and get an answer grounded in the indexed code. Feature Mode goes the other direction, letting you plan new architecture against the map before writing any of it.

Nothing else we evaluated keeps a persistent, repo-wide picture. Per-PR diagrams get thrown away after the merge, which means the tool relearns your system on every change and you never get to build on it. More on code navigation, and it is the piece we would most like you to poke holes in in the launch thread.

Analytics you can ask questions of

Review latency, rework after findings, merge throughput and DORA metrics, per team and per repository, from the day you connect a provider. Underneath that: impact scoring, bus factor, work-distribution fairness, and burnout risk per contributor.

The uncomfortable ones are deliberate. Bus factor and workload distribution are the metrics that predict which team is about to have a bad quarter, and they are almost never on the dashboard. More on developer analytics.

Monitoring, and a CLI

Codity watches repositories continuously and surfaces anomalies in review and deployment patterns to Slack, email or your tracker, ahead of the incident.

And all of it runs locally. codity with no arguments gives you a home screen that already knows which repository you are standing in and what has changed in it. Review, scan and risk-analysis run from the terminal and in CI, and there is a Claude Code plugin for the people who live in an agent. More on the CLI.

Where it runs

GitHub, GitLab, Bitbucket and Azure DevOps. All four, not one with three on a roadmap.

That mattered to us because the teams with the hardest review problems are rarely the ones on a single tidy GitHub org. They are the ones with a decade of history spread across providers, which is also exactly the situation the architecture map is for.

Merge gating, for the teams that need it

Policy checks on reviewers, linked tickets, tests and required sections, enforced as a hard merge-blocking status check. Overrides need a second person and are written to an audit trail. If you have ever had to produce evidence for SOC 2 about who bypassed what, you know why that is a feature and not a formality.

Try it

You can start on your own repositories in a few minutes: connect a provider, let the index build, open a pull request. The pricing page has the plans, and if you want the guided version, book a demo and we will walk your codebase with you.

If Codity is useful to you, an upvote and a comment on Product Hunt genuinely helps. The comments are where we get the sharpest feedback, so tell us what is missing.

Thank you

To the teams who ran early versions on real pull requests and told us plainly when a finding was noise: this product is shaped by those conversations more than by anything we planned. Thank you.

If you want to be part of the next round of that, the Product Hunt thread is where we are answering everything today.

We are not here to predict the future of code quality. We are here to build it. Today is one step of that, and we are glad you are here for it.