What is Macroscope
Macroscope is an AI agent designed to analyze codebases. It provides summaries of code activity, detects bugs, and reviews pull requests.
How to use Macroscope
- Connect GitHub: Install the GitHub app. Macroscope will automatically start summarizing commits and pull requests, identifying projects, providing productivity statistics, and performing code reviews on PRs.
- Connect Jira / Linear: Integrate with your ticketing system (Linear or Jira) to enrich Macroscope's understanding of project context and work drivers. This step is optional.
- Connect Slack: Integrate with Slack to ask Macroscope questions directly and receive answers grounded in your codebase. Macroscope researches your codebase, git logs, and issue tracking systems to provide these answers.
Features of Macroscope
- Codebase Analysis: Understand codebase changes and activity.
- Summaries: Generates summaries of commits and project activity.
- Bug Detection: Identifies bugs and potential issues in code.
- PR Review: Provides automated code reviews for pull requests.
- Engineering Time Allocation: Visualizes how engineering time is spent.
- Contributor Summaries: Summarizes the work done by individual contributors.
- Code Questions: Answers questions about the codebase, including model usage and credential management.
- Slack Integration: Allows users to interact with Macroscope via Slack for code-related queries.
Use Cases of Macroscope
- For Leaders: Gain visibility into product development, understand codebase changes, and track engineering time allocation.
- For Engineers: Save time on reporting and manual summaries, focus on coding, get automated PR reviews, and quickly find answers to code-related questions.
- Code Review Enhancement: Improve the quality and speed of code reviews with AI-generated descriptions and bug detection.
Pricing
Macroscope offers a "Teams" plan priced at $30 per user per month. A calculator is available to estimate savings.
FAQ
- What models does Macroscope use for code review? Macroscope uses OpenAI's
o4-mini-highfor an initial pass and Anthropic'sOpus 4for consensus on correctness issues to reduce false positives. - Why did Macroscope switch to JWTs? To unify and secure credential validation across services, reduce token misuse surface area, enable stateless auth, and simplify auditing by embedding scopes and user IDs in tokens.
- Does Macroscope train models on customer code? No, Macroscope does not train models using customer source code, and agreements with model providers ensure they also do not train on customer IP.
- How is customer data secured? Customer data is encrypted at rest and in transit. Code is architecturally isolated and secured, with no employee access to source code.
