Scout
An engineering agent that investigates issues across large codebases.Scout indexes codebases, logs and tickets. It traces bugs to their root cause, proposes fixes with tests, and opens pull requests for engineers to review.
- issue: invoice tax totals off by ¥1 for some customers
- ✓ logs412 mismatches · all mixed 8% / 10% rates
- ✓ tracebilling → tax-service → round()
- ✓ reproduced in sandboxrounded per line, not per rate
- ✓ patch + regression test3 files · tests passing
- PR #1182 openedreviewer: @billing-team
- design-partner engineering teams
- 4
- design-partner engineering teams
- lines of code indexed (largest repo)
- 3.2M
- lines of code indexed (largest repo)
- of beta PRs merged with minor or no edits
- 31%
- of beta PRs merged with minor or no edits
- general availability planned
- Beta
- general availability planned
Overview
Why we built Scout.
Much engineering time goes into understanding rather than writing code: finding where something happens, why it broke and what a change might affect. Scout builds a map of the codebase, including its structure, call graphs and history, and connects it to logs, traces and tickets.
Given an issue, Scout investigates like a senior engineer would. It reproduces the problem, forms hypotheses, checks them against code and telemetry, and opens a pull request with a fix, a regression test and an explanation of the root cause. Engineers review and merge; Scout never pushes directly.
Capabilities
Core capabilities.
Codebase map
Symbol graph, call paths, ownership and change history, kept up to date on every merge.
Issue investigation
Connects tickets, logs and traces to the code paths involved.
Fix proposals
Pull requests with a patch, a regression test and a root-cause write-up.
Sandboxed execution
Runs builds and tests in isolated environments, never against production.
Architecture
System architecture.
A simplified view of the principal layers and components.
Built with
Sources
GitIssue trackerLogs & tracesCIIndex
Symbol graphEmbeddingsHistoryOwnershipAgent
Investigation loopSandbox runnerTest generationOutput
Pull requestsRoot-cause notesSlack summaries
Contact
Have an AI product in mind?
Share what you would like to build, and we will help turn the idea into a production-ready product.
No technical specification required.
What happens next
- 1
Share your idea
Describe the problem, the users and the outcome you need.
- 2
Discovery call
We review your requirements together and clarify scope and constraints.
- 3
Proposal
You receive a clear scope, technical approach and delivery plan.