Ship features, fix bugs, triage issues, write tests, and generate documentation — all autonomously, inside your existing stack.
BA interview → spec → AI coding in Docker → GitLab MR. The full cycle from business language to reviewed, deployable code.
AI triages queued bugs in batch, assesses severity and root cause, then fixes auto-fixable issues autonomously — each with a regression-tested MR.
Checks new MRs for missing tests and writes them automatically. Backfills historic gaps to keep coverage high across your entire project.
AI reviews every merge request before it merges — catching missing tests, N+1 queries, unbounded loads, and code smells before your team does.
AI reads your codebase and produces user guides, developer docs, and API references — kept in sync as code evolves.
Ask questions about any project in plain English. Koderift reasons over your live codebase to answer architecture, ownership, and implementation questions instantly.
Automatically patches vulnerable dependencies when CVEs are published. Updates, tests, and opens an MR — before you even see the advisory.
Business analysts describe the feature in plain language. Koderift conducts a structured AI interview, probing for edge cases, acceptance criteria, and constraints.
Natural language → requirementsThe platform fetches your live codebase context on demand. A precise technical spec is generated and routed to dev for sign-off.
File-on-demand contextKoderift's AI agent runs in an ephemeral Docker container — isolated network, matching runtime, private gems resolved. It writes, tests, and refines without babysitting.
Isolated Docker runnerAn MR lands in GitLab with a full implementation summary and test results. Developers review, approve, and merge — no onboarding, no context loss.
GitLab MR → staging → prodBugs flow in from Sentry webhooks and manual reports. Koderift pulls the stack trace, error context, and affected endpoints automatically — with source-aware context for each.
Sentry webhook Manual reportsThe AI evaluates severity, infers root cause, identifies affected files, and routes each bug: auto-fixable issues go straight to the fix pipeline; complex bugs get flagged for a developer interview.
Root cause analysis Smart routingFor auto-fixable bugs, Koderift branches from staging, writes a targeted fix in an isolated Docker environment, and generates regression tests to prevent recurrence.
Isolated environment Regression testsA merge request lands in GitLab with the fix, regression test, and a full explanation. Your team reviews the diff — the fix is ready before they finish coffee.
Auto MR Regression tests includedWhen a new merge request lands, Koderift scans the changed files for missing test coverage — untested methods, uncovered branches, and new code paths without assertions.
Automatic on every MRThe AI agent understands your testing patterns, fixtures, and conventions. It writes targeted tests that match your project's style — not generic boilerplate, real coverage.
Matches your test style Real assertionsBeyond new MRs, Koderift can scan your entire project for untested code and backfill missing tests in batches — fixing years of coverage debt incrementally.
Project-wide backfill Incremental batchesMR Guard fires automatically whenever a merge request is opened or updated in GitLab. No configuration per MR — it watches the whole project.
Automatic trigger GitLab integrationThe AI reads the full diff alongside the original feature spec. It flags missing tests, N+1 queries, unbounded database loads, security concerns, and logic errors — with the original intent as context.
Spec-aware review Performance checksReview comments land directly on the MR in GitLab. Blockers keep the MR in draft; on a clean pass MR Guard approves and undrafts — ready for a human reviewer.
Inline MR comments Auto-undraft on passEvery MR gets the same rigorous review regardless of who wrote it. Human reviewers spend time on architecture and business logic — not spotting missing indexes.
Consistent quality barType any question about a project — architecture, ownership, implementation details, or where a given piece of behaviour lives. No query language, no setup. Just ask.
Natural language Any projectKoderift fetches live codebase context on demand — reading relevant files, tracing call paths, and reasoning across models, controllers, services, and specs to build a precise answer.
Live codebase context Cross-file reasoningYou get a clear, cited answer pointing to the exact files and line ranges. Onboard new developers in minutes, answer stakeholder questions instantly, and debug faster.
File citations Instant onboardingKoderift analyses your repository structure, models, controllers, services, and API endpoints to understand what needs documenting and how it all connects.
Automatic discoveryUser guides for end users, developer documentation for your team, and complete API references — each tailored to its audience with the right level of detail.
User guides Dev docs API refsAn AI-guided interview lets you refine tone, scope, and focus areas. Upload screenshots, add context, and shape the documentation to match your team's standards.
Guided interviewDocumentation is versioned with your codebase. When code changes, Koderift detects drift and regenerates affected sections — your docs never go stale.
Auto-sync on code changesKoderift monitors your project dependencies against published CVEs — whether they're Ruby gems, npm packages, or any other package manager. When a vulnerability is found, it identifies the affected package, severity, and the safe version to upgrade to.
Continuous monitoring Severity assessmentThe vulnerable package is updated to the patched version in an isolated environment. Koderift resolves dependency conflicts, updates the lockfile, and verifies compatibility with your stack.
Dependency resolution Lockfile updateYour full test suite runs against the updated dependency. If everything passes, an MR is opened with the CVE details, upgrade diff, and test results — ready for a quick review and merge.
Full test suite CVE details in MRFeatures, bug fixes, MR reviews, test coverage, project intelligence, and documentation — all autonomous, all inside your stack.