Pull Request Explorer

Summary

Add command to stage coarser repartitions

Adds a management command that stages coarser datetime partitions for data warehouse tables, enabling operators to efficiently roll back over‑partitioned tables without manual admin intervention.

Health Assessment

• • Quick review and merge with minimal iterations suggest a well‑tested change; the addition of a new command provides operational benefit with low risk.

Summary

Fix growth labeler pacing and deferral handling

Improves AI budget utilization and reliability by pacing web tool calls, preventing unnecessary deferrals, and reducing alert noise, ensuring cost-effective enrichment runs.

Health Assessment

• • Rapid 4‑hour cycle with minimal review friction; AI assistance streamlined code changes; comprehensive tests added to guard against regressions.

Summary

Fast attribute keys when no metric selected

Optimizes metrics filter performance by querying pre-aggregated attribute keys when no metric is selected, cutting latency from seconds to ~1 second.

Health Assessment

• • Rapid review and merge within 1.7 hours, minimal rework, AI-assisted code contributed to performance improvement.

Summary

Close telemetry gaps for suggested scouts strip

Adds analytics events and suggestion bookkeeping to track user interaction with the scouts strip, improving adoption measurement.

Health Assessment

• • Fast 2.2‑hour cycle with minimal rework; AI‑assisted code generation; feature behind a flag; low risk to production.

Summary

Move Score Now to Background Inference Workflow

Reduces browser timeouts and duplicate runs by executing scoring asynchronously, improving reliability and user experience.

Health Assessment

• • Fast cycle time and single review round indicate efficient process, but large code changes suggest significant impact.

Summary

Fix CI: post depot gate check in one request

Ensures the Depot gate check posts verdict and output in a single request, preventing loss of output and improving CI status reliability.

Health Assessment

• • Rapid review and merge with minimal changes to CI workflow, indicating low complexity and risk.

Summary

Health Assessment

Summary

Fix validation retry limits under memory pressure

Improves stability of Cloud Claude agents by limiting validation retries under memory pressure, reducing unnecessary restarts and resource consumption.

Health Assessment

• • Quick turnaround with minimal review comments, indicating a straightforward bug fix.

Summary

Skip log production when logger loop closed

Prevents runtime errors in Temporal activities by guarding against closed event loops, ensuring log lines don’t cause failures.

Health Assessment

• • Quick fix with minimal changes and immediate approval indicates low complexity and high confidence.

Summary

Retry depot backend tests on GitHub re-run

Improves developer experience by ensuring failed depot tests are retried automatically when a GitHub re-run is triggered, eliminating manual steps. This reduces friction and speeds up debugging.

Health Assessment

• • Quick turnaround with minimal rework; single review and merge within 2 hours indicates high confidence and low risk.

Summary

Add project and personal agent instructions for cloud runs

Enables users and admins to store custom instructions for PostHog cloud agent runs, improving automation and personalization.

Health Assessment

• • The PR was completed in under six hours with a single review, indicating efficient collaboration. The large code changes across backend and frontend suggest significant impact, but the quick review cycle and AI assistance helped mitigate potential friction.

Summary

Add row checkboxes for space feed sessions

Adds Gmail‑style checkboxes to space feed rows, enabling bulk actions like pin, archive, and file to space, improving productivity for users. The feature is behind a flag and includes comprehensive tests.

Health Assessment

• • Fast cycle time and immediate approval indicate high confidence and minimal review friction. Despite the large scope, the PR was delivered with no blockers and extensive testing.

Summary

Add task source display to hover card and feed

Users can now see which product created a task directly in the UI, improving context and reducing confusion.

Health Assessment

• • Fast cycle time and single review round indicate straightforward UI change with minimal risk.

Summary

Add living documents view to artifacts tab

Users can now see and preview living documents such as Slack canvases directly in the Artifacts tab, improving visibility and reducing download friction.

Health Assessment

• • Fast cycle time (1.2h) and single review round indicate a smooth, low‑friction process with minimal rework.

Summary

Fix silent failure of metrics characterization

Prevent error toast when metrics characterization fails, improving user experience.

Health Assessment

• • Quick fix with minimal changes and a regression test, approved within minutes.

Summary

Fix duplicate DMARC record in email DNS

Prevents email domain configuration from creating duplicate DMARC records, ensuring existing policies remain effective.

Health Assessment

• • Quick AI-assisted fix with minimal rework; fast review and merge cycle, low risk to production

Summary

Migrate activity descriptions to product folder

Reorganizes experiment activity description code to improve developer workflow and reduce merge queue delays.

Health Assessment

• • Quick merge with minimal review, mechanical changes, AI-assisted authoring.

Summary

Stabilize overview table height

Prevents UI flicker during table refresh, improving user experience and reducing confusion.

Health Assessment

• • Quick approval with minimal changes indicates low complexity and high confidence.

Summary

Fix Cloudinary source response for data imports

Ensures Cloudinary syncs correctly by wrapping REST resource in SourceResponse, preventing import failures and enabling proper partitioning. This resolves AttributeError during syncs and improves reliability.

Health Assessment

• • Quick fix with minimal changes and immediate approval indicates low complexity and high confidence in the patch.

Summary

Add theme-aware tooltip for experiment results

Users now see screenshots that match the page theme, improving clarity and reducing confusion in light and dark modes.

Health Assessment

• • The PR was reviewed and merged within 2 hours with minimal rework, indicating a smooth process and low risk for the UI change.

Summary

Add view list, edit review, goal results to broadcasts

Users can now view recipient lists, edit steps directly from the review screen, and see goal conversion metrics on broadcasts. This improves workflow efficiency and provides clearer performance insights.

Health Assessment

• • Fast review and merge, minimal rework, straightforward UI enhancements.

Summary

Fix ingestion shadow read comparison for identity

This change corrects shadow mode read comparison to avoid false divergence metrics, ensuring accurate parity checks. It reduces unnecessary alerts and improves data consistency monitoring.

Health Assessment

• • Quick review and merge with minimal changes, indicating low complexity and risk.

Summary

Add empty space feed welcome with suggestions

Introduces a welcome screen for empty space feeds, offering desktop‑style suggestions that pre‑populate the composer, enhancing user onboarding and reducing friction.

Health Assessment

• • Fast 1.4‑hour cycle, single commit, zero review lag, AI‑generated code, minimal scope, low risk

Summary

Make daily briefing shorter and easier to scan

Shortens the daily briefing content to improve readability and scanning speed for users.

Health Assessment

• • Fast review and merge with minimal changes, low risk.

Summary

Fix latest prediction batch selection

Corrects the predictions view to display the full latest batch by selecting on prediction date, improving data accuracy for users.

Health Assessment

• • Fast turnaround with a single commit and immediate approval indicates low complexity and minimal risk.

Summary

Add source guidance section to self-driving docs

Adds a new section explaining how to steer sources using labels and MCP, giving users better control over automated issue handling.

Health Assessment

• • Fast review and merge with minimal changes; AI-assisted documentation update.

Summary

Fix funnel chart white-screen on metric swap

Prevents blank page when funnel metrics change, improving reliability for experiment results.

Health Assessment

• • Quick fix with tests, minimal changes, fast review, no blockers.

Summary

Avoid uv exit segfault in backend CI

Prevents random CI failures caused by uv thread detachment on Ubuntu 24.04, ensuring reliable CI runs.

Health Assessment

• • Fast turnaround with minimal changes, no user-facing impact, and a single review cycle.

Summary

Fix secret-field guard to scan entire repo

Ensures secret fields are properly detected across the repository, preventing accidental exposure of sensitive data.

Health Assessment

• • Quick AI-assisted fix with minimal code changes, fast review turnaround, and low risk to production.

Summary

Show cited PostHog objects live in artifacts

Enables users to view referenced insights, dashboards, and replays directly within task artifacts, improving visibility and reducing navigation friction.

Health Assessment

• • The PR was merged within 2.3 hours with a single review cycle, indicating a smooth review process. The moderate code change size and frontend focus suggest low risk to production.

Summary

Add copy link and video preview for artifacts

Users can now copy artifact URLs and view video artifacts directly in the UI, improving workflow efficiency.

Health Assessment

• • Rapid approval with minimal review, indicating a straightforward change.

Summary

Fix merge-queue baseline drop failures

Ensures merge-queue runs fail when baseline entries are missing, preventing silent deletions and protecting visual review integrity.

Health Assessment

• • Rapid AI-assisted review enabled a quick fix for baseline integrity, keeping visual review reliable.

Summary

Add case-control sampling for large training populations

Enables autoresearch training on populations exceeding the row cap by sampling positives and negatives, preserving accuracy while scaling.

Health Assessment

• • Fast cycle time with minimal review indicates high confidence; large scope confined to autoresearch module suggests manageable risk.

Summary

Fix cloud Codex startup on missing MCP server

Ensures cloud Codex tasks fail loudly when the PostHog MCP server is unavailable, preventing silent failures and improving reliability.

Health Assessment

• • Rapid approval with minimal changes indicates low complexity and high confidence in the fix.

Summary

Align refresh button timestamp with label

Fixes visual misalignment in experiments UI, improving readability and consistency.

Health Assessment

• • Quick fix with minimal changes, approved swiftly, no major risk.

Summary

Show session files on space feed cards

Adds UI chips to display agent‑produced files in space feed cards, improving visibility of artifacts for users.

Health Assessment

• • The PR was merged within 48 minutes of opening, received a single approval with no rework, indicating a smooth review process for a small, frontend‑only change behind a feature flag.

Summary

chore(deps): bump virtualenv from 20.36.1 to 21.7.12

Updates the virtualenv dependency to the latest version, incorporating security and bug fixes for the Python environment.

Health Assessment

• • Automated dependency bump with a single file change, merged quickly with minimal review effort.

Summary

Bump Django dependency to 5.2.17

Updates Django to the latest patch, addressing security fixes and bug patches.

Health Assessment

• • Dependency bump merged quickly with minimal changes, indicating efficient automated process.

Summary

Update uv dependency across multiple directories

Bumps urllib3 and related packages to the latest secure versions, ensuring all example projects stay up‑to‑date and compatible.

Health Assessment

• • Quick dependency update with no review needed, minimal risk.

Summary

Show session creator face and adjust row weight

Adds presence badges to the Today rail, improving visibility of active collaborators and aligning the web UI with Desktop behavior.

Health Assessment

• • The PR was reviewed and merged within 0.8 hours, indicating a smooth, low‑friction process with minimal rework.