DOCSEARCHbyAlgolia
Submit Docs

Aviator Documentation

website·Indexed 35 days ago

https://docs.aviator.co/

Aviator is an AI-coding platform featuring Runbooks, which use Claude Code agents to plan and execute coding tasks in isolated sandboxes. It includes a developer dashboard called Inbox for managing GitHub pull request reviews and tracking workflows, as well as tools for automated code refactoring, bug fixes, migrations, and improving code coverage.

Tokens
65.3K
Snippets
159
Records
442
Agent score
65%

What's inside Aviator

  1. Overview of Aviator Inbox

    Aviator Inbox is a personalized dashboard for managing GitHub pull request (PR) reviews. It replaces the previous AttentionSet feature and organizes PRs into sections based on the user's relationship to the PR and its current state. The dashboard updates in real-time as GitHub events occur (e.g., comments, CI failures, or approvals).

  2. Overview of Aviator One shot mode

    One shot mode is an automated execution mode for Runbooks that streamlines the entire code modification workflow into a single, uninterrupted process. Unlike standard mode, it skips clarifying questions and uses reasonable defaults to move from task description to a completed Pull Request automatically.

  3. Overview of SSH sandboxes for Runbooks

    SSH sandboxes provide a self-managed execution environment for Runbooks, allowing organizations to run Claude Code on their own infrastructure instead of using Aviator's managed cloud sandboxes. This is primarily used for enhanced security, enterprise compliance, and cost management by eliminating per-execution charges. The system connects to remote Linux environments via SSH transport, using key-based authentication and encrypted private key storage.

  4. Overview of Aviator Verify

    Aviator Verify shifts code review from analyzing diffs to verifying that a change achieves its intended purpose. It captures developer intent via a CLI call, runs acceptance criteria against the code, and generates a review document containing verdicts and evidence for the reviewer to judge.

  5. Overview of Runbooks for AI-driven coding

    Runbooks uses Claude Code agents to plan and execute coding tasks within isolated sandboxes. It enables a spec-driven workflow where users describe a desired outcome, review a generated implementation plan, and allow agents to execute the steps. Key features include Multiplayer AI for team collaboration in shared sessions, persistent knowledge via context files and templates, and a structured execution process to ensure consistency across teams.

  6. Overview of On-Premises SSH Sandboxes

    SSH Sandboxes allow Claude Code to execute runbook steps on your own servers or virtual machines. This is recommended for users who need access to private internal resources, have strict security/compliance requirements, want full control over the execution environment and dependencies, or are working with on-premises infrastructure.

  7. Overview of automated code refactoring with Runbooks

    Runbooks provides automated code refactoring by analyzing a codebase, identifying improvement opportunities, and systematically updating code to modern patterns. This is particularly useful for legacy code modernization, framework upgrades, performance optimization, and standardizing architectural patterns across large codebases.

  8. Overview of Aviator remote agent workflow

    The remote agent flow allows you to run coding tasks in an Aviator-managed sandbox using Runbooks, rather than working locally in your editor. This is ideal for batch processing multiple tasks, off-hours work, or tasks driven by non-developers (e.g., PMs or security teams) who can file ticket-style tasks. While the implementation step moves to a remote sandbox, the MCP tools, verification process, and resulting review documents remain identical to the local flow.

  9. Overview of Aviator Releases

    Aviator Releases is a unified dashboard designed to help product engineers manage releases, deployments, cherry-picks, and rollbacks across multiple services and environments. It aims to democratize the deployment process by providing visibility and guardrails, reducing reliance on a small set of specialized engineers and minimizing human error during brittle deployment processes.

    It integrates with GitHub and the following CI/CD systems:

    • Jenkins
    • Buildkite
    • CircleCI
    • GitHub Actions
    • ArgoCD
    • Spinnaker
    • FluxCD

  10. Overview of Batch Operations with Runbook Templates

    Batch operations allow you to apply a single Runbook template across multiple targets (up to 200 repositories or directories) simultaneously. The system creates individual "fanouts" for each target, which are processed concurrently with automatic retry and timeout handling. This is ideal for dependency upgrades, code migrations, security patches, and enforcing style consistency across an organization or monorepo.

  11. Overview of Aviator's AI code shipping suite

    Aviator provides a suite of building blocks to handle the process after AI generates code, focusing on capturing intent, verifying behavior, queuing merges, coordinating releases, and managing reviewer attention.

  12. Understand when Verify triggers preview environments

    Previews are ephemeral environments that only spin up when a verification run contains at least one runtime criterion—a check that the classifier determines must be executed against running code. If all criteria can be verified via code-scan alone, no preview is built, reducing costs and increasing run speed.