Autter.dev

Code-aware engineering intelligence, from pull request to production

Understand the codebase. Review the change. Verify the risk. Learn from runtime.

Autter connects repository structure, pull-request history, review findings, and runtime evidence so engineering teams can decide what is safe to merge and what needs attention next.

Install the GitHub App and Autter builds a living model of your codebase: services, symbols, dependencies, ownership, APIs, schemas, hotspots, and change history. That context powers PR review, codebase scans, generated documentation, incident investigation, and engineering automation.

AI coding tools increased the volume and speed of code production. Reviewers still need to understand how a change affects the rest of the system, whether a migration can damage data, and whether code that looks plausible will work in production.

Diff-only review misses those questions. Autter reads the change against the codebase around it, checks the evidence with focused analyzers, and gives maintainers a clear merge state with the findings behind it.

Pull request
     │
     ▼
Risk and reviewability assessment
     │
     ▼
Repository-aware review
  ├── changed files and full-file context
  ├── dependencies and downstream consumers
  ├── semantic siblings and co-change history
  └── repository rules and prior learnings
     │
     ▼
Adversarial verification and deduplication
     │
     ▼
Inline findings + final task list + merge gate
     │
     ▼
New commits trigger a fresh decision

Autter adjusts the depth of review to the size and risk of each change. A second pass classifies candidate findings as confirmed, uncertain, or false positive. Developers see the useful results in GitHub, while maintainers can tune sensitivity, verbosity, draft behavior, bot-authored PR policy, ignored paths, and check families.

Run scans on demand, on a schedule, after a default-branch push, or when you tag a release. Autter combines focused security and quality agents with established tools for SAST, dependency analysis, containers, infrastructure as code, API exposure, licenses, secrets, and software supply-chain risk.

The report ranks and consolidates findings. Reachability, exploit evidence, severity, and repository context help your team decide what to fix first.

Ask about a symbol, its callers, a schema, a previous review, or the likely blast radius of a change. Autter searches source, symbols, scopes, findings, learnings, and generated documentation. Answers include citations back to the evidence.

Autter generates a versioned repository wiki with architecture, API, data-model, onboarding, ownership, module, and technical-debt pages. Incremental updates refresh affected documentation as the code changes.

Send telemetry through OpenTelemetry or the Autter Node SDK. You can also connect Sentry, PostHog, Grafana, or a normalized webhook source. Autter groups occurrences into incidents, links them to deployments and pull requests, and investigates the relevant code. For qualified cases, Captain Patch can prepare a bounded fix PR after verification and safety checks.

Autter records the evidence behind each review and preserves repository-specific learnings for future changes. Teams can measure finding acceptance, fix rates, review speed, severity trends, engineering health, and line-level AI authorship.

Custom agents extend the same context to work such as reviewer routing, release notes, test-gap checks, migration policy, dependency audits, documentation drift, and scheduled engineering summaries.

Each Autter organization has independent access controls, organization-scoped artifact storage, and a dedicated PostgreSQL database for operational data. Analysis jobs can use disposable databases and isolated execution environments. Runtime credentials stay scoped to one repository; Autter displays raw keys once and stores their hashes.

• GitHub: Reviews, inline findings, checks, merge state, issues, and pull requests
• Web app: Repository intelligence, scans, wiki, runtime, analytics, and administration
• CLI: Device-authorized access, AI authorship metadata, and provenance-aware blame
• MCP: Codebase queries, impact analysis, reviews, scans, dependency checks, and repository health inside compatible coding tools
• Slack and email: Alerts and custom-agent outcomes

Engineering leads use Autter to keep review quality steady as AI-assisted output grows.

Platform and security teams use it to apply repository policy, scan codebases, trace change impact, and manage risk across many repositories.

Maintainers use it to spend less time reconstructing context and more time deciding whether a change belongs in the codebase.

Connect GitHub, choose a repository, and let Autter build its index. A guided setup walks you through runtime instrumentation, wiki generation, codebase scans, and team access. You can also create a demo repository with planted findings before connecting production code.

Get started at autter.dev

• Product Handbook for the product, architecture, and operating principles
• GitHub Discussions for build updates, ideas, and Q&A
• The Harbour Logs for notes from the founders
• Brand Story for the origin of Autter and Captain Patch

Autter is moving fast, and some repositories may not accept external code contributions yet. You can still help by opening an issue, joining a discussion, or testing Autter on a repository and sharing the misses.

See CONTRIBUTING.md for the current contribution policy.

Visit autter.dev, start a discussion, or email hi@autter.dev.

_{Captain Patch, Autter's code-aware engineering agent.}