Public repo examples

Four PRs that passed review and shouldn't have.

Real public pull requests in Java, Python, TypeScript, and C#. Every one of them looked routine in the file diff, got approved, and quietly moved a boundary somebody will pay for later. Open any of them on GitHub and check what Striff caught for yourself.

Open homepage gallery Install extension

What to look for

Boundary shifts, duplicated responsibilities, shared hotspots, and the relationships a file diff hides.

Why this matters

These risks are invisible in a line-by-line diff. They only show up when you look at the change as a dependency graph, the way Striff does on every PR.

The reviews

What a reviewer would have missed.

Each example shows the architecture diagram Striff generated and the GitHub Check it posted, exactly what your team would see on their own PRs.

Java apache/pinot #19073 Public PR

Generate Avro schemas from Pinot logical data types

7 findings · 9 components · HIGH risk

Open Generate Avro schemas from Pinot logical data types

apache/pinot #19073

Conversation 22 Checks 11 Files changed 17

striff / architectural-review

Architectural review completed

Completed with 3 findings across 9 components in the structural diff.

Java Striff structural diff for apache/pinot #19073

striff / architectural-review

Completed in GitHub Checks

Completed 1 check run with findings

Summary Findings 3

Review headline

The core engine took its first dependency on a plugin.

Striff flagged the inverted plugin boundary before merge, with the first core-to-plugin edge named in the check.

9 components in the structural diff 7 findings including 2 HIGH severity

New directional boundary crossing

New dependency from org.apache.pinot.core.util to org.apache.pinot.plugin.inputformat.avro, with no prior edge in this direction.

Stable contract modified: 24 dependents

AvroUtils has afferent coupling of 24. Changes to this component impact at least 24 dependents in the parsed scope.

Layer skip on the new edge

The new core-to-plugin edge jumps from core.util (layer 1) to plugin.inputformat.avro (layer 3), skipping a layer.

This PR teaches Pinot to generate Avro schemas from logical data types. Striff caught the direction of the change: org.apache.pinot.core.util now depends on org.apache.pinot.plugin.inputformat.avro, the first edge ever from the core into the Avro plugin package. Plugin boundaries exist so the core never knows about plugins; this PR quietly reversed that.

Open PR on GitHub

Python celery/celery #10418 Public PR

Preserve absolute event timestamps in Gossip

4 findings · 7 components · HIGH risk

Open Preserve absolute event timestamps in Gossip

celery/celery #10418

Conversation 22 Checks 11 Files changed 17

striff / architectural-review

Architectural review completed

Completed with 3 findings across 7 components in the structural diff.

Python Striff structural diff for celery/celery #10418

striff / architectural-review

Completed in GitHub Checks

Completed 1 check run with findings

Summary Findings 3

Review headline

A timestamp fix that wired the smoke suite four layers deep into Celery's core.

Striff flagged both new boundary crossings and the 3-layer skip before merge.

7 components in the structural diff 4 findings including 2 HIGH severity

New directional boundary crossing

New dependency from t.smoke.tests to celery.utils, with no prior edge in this direction.

New directional boundary crossing

New dependency from t.smoke.tests to celery.worker.consumer, with no prior edge in this direction.

Layer skip: 3 layers

The new edge from t.smoke.tests (layer 0) to celery.utils (layer 4) skips 3 layers of the codebase's layering.

This PR stops Gossip from losing absolute event timestamps. The fix is fine. The new smoke tests are the story: t.smoke.tests now imports celery.utils and celery.worker.consumer directly, two dependency directions that never existed before, one of them jumping from the test layer straight to a utility layer four levels down.

Open PR on GitHub

TypeScript typeorm/typeorm #12647 Public PR

fix: remove require() calls that break bundlers

4 findings · 7 components · HIGH risk

Open fix: remove require() calls that break bundlers

typeorm/typeorm #12647

Conversation 22 Checks 11 Files changed 17

striff / architectural-review

Architectural review completed

Completed with 3 findings across 7 components in the structural diff.

TypeScript Striff structural diff for typeorm/typeorm #12647

striff / architectural-review

Completed in GitHub Checks

Completed 1 check run with findings

Summary Findings 3

Review headline

A bundler fix that rewired two mobile drivers into the platform layer.

Striff flagged both new driver-to-platform crossings and the layer skip on each of them, before merge.

7 components in the structural diff 4 findings including 2 HIGH severity

New directional boundary crossing

New dependency from src.driver.react-native to src.platform, with no prior edge in this direction.

New directional boundary crossing

New dependency from src.driver.expo to src.platform, with no prior edge in this direction.

Layer skip on both new edges

Both edges jump from the driver layer (layer 1) straight to src.platform (layer 3), bypassing the intermediate layer.

This PR removes require() calls so bundlers can statically analyze TypeORM. Striff caught the structural side effect: the React Native and Expo drivers now depend directly on src.platform, two boundary crossings with no prior edge in that direction, each skipping a layer.

Open PR on GitHub

C# dotnet/efcore #38676 Public PR

Handle out-of-order split include rows during concurrent inserts

8 findings · 11 components · HIGH risk

Open Handle out-of-order split include rows during concurrent inserts

dotnet/efcore #38676

Conversation 22 Checks 11 Files changed 17

striff / architectural-review

Architectural review completed

Completed with 3 findings across 11 components in the structural diff.

C# Striff structural diff for dotnet/efcore #38676

striff / architectural-review

Completed in GitHub Checks

Completed 1 check run with findings

Summary Findings 3

Review headline

A concurrency fix that pointed the query internals back at the root namespace.

Striff flagged the new upward edge, the 3-layer skip, and the near-cycle it left behind, before merge.

11 components in the structural diff 8 findings including 1 HIGH severity

New directional boundary crossing

New dependency from Microsoft.EntityFrameworkCore.Query.Internal to Microsoft.EntityFrameworkCore, with no prior edge in this direction.

Layer skip: 3 layers

The new edge jumps from Query.Internal (layer 8) straight to the root namespace (layer 12), skipping 3 layers.

Cyclic dependency seed

The new edge, combined with the existing path from the root through Query to Query.Internal, almost creates a namespace cycle.

This PR fixes split-include queries dropping child rows under concurrent inserts. Striff caught the structure: SplitQueryResultCoordinator in Query.Internal now depends on the root Microsoft.EntityFrameworkCore namespace, where the PR's new DbQueryConcurrencyException lives. That direction never existed before, it skips 3 layers, and it leaves the query namespaces one step from a full cycle.

Open PR on GitHub