Pull Request Explorer

Summary

Centralize experiment exposure event contract

Standardizes exposure event handling across experiments, reducing hardcoded literals and future‑proofing for upcoming flag config changes.

Health Assessment

• • Quick turnaround with minimal rework; refactor completed within 4 hours, indicating efficient collaboration.

Summary

Fix scout scope derivation in orphan-run reconciler

Ensures scout runs retain full write permissions after re-dispatch, preventing incomplete investigations and improving reliability for user-facing reporting features.

Health Assessment

• • The PR had a single review round with AI-assisted feedback, and the small code change resolved a critical permission bug without introducing new risk.

Summary

Update quarterly account plan template to opp-focused format

Provides a leaner, opportunity‑focused quarterly account plan template, improving alignment and clarity for account managers.

Health Assessment

• • Fast review cycle (1.3h to first review) and minimal changes (12 lines added, 26 removed) indicate a straightforward documentation update with low risk.

Summary

Drop tool domains from exec command reference

Reduces serialized size to keep the exec tool in the connector registry, improving compatibility with Claude web/desktop.

Health Assessment

• • Quick fix with minimal changes, approved in under 15 minutes, no major risk.

Summary

Rework em-dash guidance into user-facing copy

Improves copy consistency by replacing brittle em-dash rule with comprehensive user-facing copy guidelines, enhancing readability across UI and documentation.

Health Assessment

• • Quick review and merge with minimal changes, indicating straightforward documentation update.

Summary

Document mobile screen event filtering via before_send

Adds documentation for filtering autocaptured screen events across mobile SDKs, enabling users to exclude unwanted screen views from event logs.

Health Assessment

• • PR merged within 30 minutes of opening, with no review comments or rework, indicating a straightforward documentation update with minimal risk.

Summary

Remove amortized credit payment from new contracts

Updates billing handbook to align with contract rules, removing unsupported amortized credit payment option for new customers.

Health Assessment

• • Quick review and merge within 30 minutes, minimal changes to documentation.

Summary

Show trace-level evaluation runs in trace view

Enables users to view all evaluation runs for a trace, improving observability and debugging of AI workflows.

Health Assessment

• • Fast review and merge (5.1h cycle, 0.3h review), minimal code changes, no blockers, low risk.

Summary

Add toIntOrDefault conversion function to HogQL

Adds integer fallback conversion to HogQL, enabling robust numeric parsing and aligning with existing float fallback, improving data integrity for AI observability signals. This fixes validation errors and supports AI-driven analytics.

Health Assessment

• • PR required several iterations to align the new function across multiple SQL dialects and tests, but AI assistance accelerated development and the final merge was achieved within 23 hours.

Summary

Refactor session replay to use element-carried recorder

Reduces coupling between replay steps and shared batch state, enabling future pipeline restructuring without affecting user-facing behavior.

Health Assessment

• • Fast 4‑hour cycle with no review friction; large code change but purely refactor, no new features or bugs introduced.

Summary

Fix drag-to-zoom single bucket selection

Improves chart zoom on sparse data, allowing users to zoom into individual buckets and trigger interval adjustments, enhancing data exploration and insight accuracy.

Health Assessment

• • Quick turnaround with a single review and merge within 2 hours indicates high confidence and low complexity.

Summary

Add blog post on helicone alternatives

Adds a new blog post comparing helicone alternatives, including a product comparison table and updated data hooks.

Health Assessment

• • PR merged within 41 hours with a single review round and minimal rework, indicating efficient collaboration.

Summary

Fix local evaluation of flag dependencies

Ensures feature flags evaluate correctly for users, preventing incorrect behavior and improving reliability across SDKs.

Health Assessment

• • The PR addressed a critical bug in flag dependency evaluation, added extensive tests, and aligned behavior with other SDKs, with a straightforward review and moderate scope.

Summary

Fix preaggregation quorum timeout to 2s

Reduces dashboard hang time from 10 minutes to 2 seconds, improving reliability and user experience.

Health Assessment

• • Quick, single-commit fix with fast review and minimal scope, indicating low risk and high confidence in stability.

Summary

Skip LLM judge evals over context window

Fixes AI observability judge errors by skipping evaluations that exceed model context window, reducing retries and error tracking noise.

Health Assessment

• • Fast review and merge with minimal rework, indicating a well‑understood bugfix and efficient collaboration.

Summary

Remove unused recalculation progress fields

This PR trims unnecessary progress data from the recalculation endpoint, reducing payload size and improving UI responsiveness.

Health Assessment

• • Fast review and merge with minimal rework indicate a well‑executed change; the moderate scope and clear business impact keep risk low.

Summary

Refactor experiment flag variant reads

Centralizes flag variant access to reduce fragility and future‑proof experiments UI.

Health Assessment

• • Quick turnaround with minimal changes and no behavior impact, indicating a low‑risk refactor.

Summary

Remove dead emits_embeddings from summarizer configs

This migration cleans legacy summarizer settings, preventing save errors and ensuring consistent data integrity for replay vision features.

Health Assessment

• • Quick fix with minimal changes, resolved in under 8 hours with a single review, indicating an efficient and low-risk process.

Summary

Harden replay player against stuck buffering

Ensures session recordings reliably play or surface errors instead of hanging, improving user experience and reducing support incidents.

Health Assessment

• • Rapid review and merge after 17.6 hours, with multiple commits addressing complex error handling, indicates efficient collaboration and high confidence in the fix.

Summary

Add stable sessions and client metadata for stateless MCP

Enables consistent session tracking and client metadata across stateless and multi-pod MCP servers, improving analytics accuracy and reducing misattributed events. This fixes a customer issue where sessions were incorrectly counted, ensuring reliable data collection.

Health Assessment

• • Fast cycle time and single review round indicate smooth integration; large code change but no rework suggests high confidence.

Summary

Add Grok logo to harness registry

Adds a Grok brand logo SVG and wires it into the harness registry so the Grok harness renders with an icon instead of a bare color slice.

Health Assessment

• • Quick review and merge, minimal code changes, AI-assisted asset addition.

Summary

Scaffold Hex import source

Adds a stub for the Hex data‑warehouse connector, providing a starting point for future development and meeting recurring user requests.

Health Assessment

• • Fast review and merge (3.8h cycle time) with minimal friction; AI-assisted code generation enabled rapid implementation of a new data‑warehouse source stub.

Summary

Isolate count-triggered eval report checks

Improves reliability of AI evaluation reports by isolating per-report checks, preventing a single slow query from failing the entire workflow and ensuring accurate report persistence.

Health Assessment

• • The PR had a moderate cycle time with a quick initial review but required multiple commits to address performance and reliability issues, indicating significant rework. The large code changes and multiple activity refactors suggest a high risk of integration impact.

Summary

Add self-driving agents repo cloning

Enables self-driving agents to fetch relevant code from additional repositories, improving accuracy of signal analysis and reducing false conclusions.

Health Assessment

• • PR had a moderate review time and multiple commits after first review, indicating iterative refinement but overall small scope and quick merge.

Summary

Add completed tone and ItemGroup combined

Adds a new purple 'completed' color token and enhances ItemGroup to support flush merged lists, improving UI clarity for completed items.

Health Assessment

• • The PR was merged within an hour of opening, with no review delays and only three commits, indicating a smooth review process.
• • The moderate number of files and lines suggests a focused change with minimal risk.

Summary

Stop importing Playwright at module level

Reduces unnecessary browser‑automation imports, speeding up backend startup and tests.

Health Assessment

• • Quick review and single commit resolved performance issue with minimal friction.

Summary

Deflake rerun-prone backend tests batch 2

Fixes flaky tests in the PostHog backend to improve test reliability and reduce rerun rates.

Health Assessment

• • The PR was reviewed and approved quickly, with no significant issues or concerns raised during the review process.

Summary

Clarify compensation levels in handbook

Updated handbook wording to clarify that project size refers to impact, improving clarity for employees.

Health Assessment

• • Quick review and merge, minimal changes to documentation.

Summary

Move insight-alerts frontend to products/alerts

Relocates insight-alerts UI components to a dedicated products folder, eliminating import cycles and improving code organization. No functional changes, purely structural refactor.

Health Assessment

• • Quick turnaround with minimal rework; the refactor was straightforward and well-received.

Summary

Support Dia browser cookies in Metabase login

Adds Dia browser support to the Metabase login CLI, enabling users to authenticate without opening Chrome and improving developer workflow.

Health Assessment

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

Summary

Fix data warehouse lease verification gaps

Ensures data pipeline reliability by preventing duplicate or lost rows during lease loss, improving consistency for downstream analytics.

Health Assessment

• • Fast turnaround with a single commit and minimal review, indicating a focused, low‑complexity fix that addresses a critical data consistency issue.

Summary

Fix Delta S3 commit safety and unsafe rename flag

Ensures safe concurrent writes to S3 by pinning conditional‑put semantics, preventing silent data corruption and making rollback explicit.

Health Assessment

• • Quick turnaround with minimal rework; AI‑assisted code and tests ensured safety without breaking existing functionality.

Summary

Retry Convex reads on mid-body connection breaks

Adds retry logic for transient network errors during Convex data imports, improving reliability of data warehouse syncs.

Health Assessment

• • Quick, single-commit fix with minimal review, demonstrating efficient resolution of a transient network issue.

Summary

Fix BigQuery missing-table error capture

Eliminates redundant error-tracking noise for BigQuery missing-table errors, improving clarity without affecting retry logic.

Health Assessment

• • Quick fix with minimal changes and single approval, indicating low complexity and high confidence.

Summary

Add PostHog CLI to products nav and ownership table

Adds PostHog CLI as a product in the navigation and ownership table, improving discoverability for terminal‑first users and clarifying product ownership.

Health Assessment

• • Quick turnaround with minimal changes; single review cycle; no major rework.

Summary

Support Codex plan mode in cloud runs

Enables Codex plan mode in cloud task API, allowing plan‑specific collaboration and handoff.

Health Assessment

• • Quick turnaround with minimal rework, indicating a straightforward change.

Summary

Fix local Modal agent overlay reliability

Ensures local Modal agent overlays correctly install missing dependencies and start only after repository readiness, reducing failures in local development environments.

Health Assessment

• • Fast cycle time and single commit indicate a focused bug fix with minimal review friction.

Summary

Add self‑improvement report escalation for custom scouts

Enables custom scouts to surface self‑improvement suggestions as inbox reports, improving visibility and actionability for owners.

Health Assessment

• • Rapid review and merge (1.1h cycle, 0.1h review) indicates low complexity and high confidence in the change.

Summary

Add token visibility and restriction shortcuts to TopHog dashboard

Admins can now view API tokens directly on the TopHog dashboard, eliminating manual lookups. The new restrictions page allows quick creation and extension of ingestion restrictions, preventing duplicate entries and improving data governance.

Health Assessment

• • Fast cycle with a single review and minimal rework, indicating clear scope and strong test coverage.

Summary

Add quill-based stat tiles and cost trend charts

Standardizes engineering analytics UI using quill components, improving consistency and maintainability for product teams.

Health Assessment

• • Fast cycle time and review, minimal rework, AI-assisted code generation streamlined the migration.

Summary

Add Rust HogVM execution for ingestion transformations

Execute transformations on the Rust HogVM behind a config flag, improving performance and reducing latency.

Health Assessment

• • The PR has a large scope with 13 files changed, but the review process was smooth with only one approved review.

Summary

Update ticket settle minutes to 1

Adjusts the ticket settlement interval to one minute, improving timing accuracy.

Health Assessment

• • Quick approval with minimal review, indicating low complexity and high confidence in the change.

Summary

Fix MCP CI cancelled job false failures

Reduces false failure rate in MCP CI by treating cancelled upstream jobs as non-failures, improving pipeline reliability.

Health Assessment

• • Quick resolution of a CI false failure issue with minimal changes, improving pipeline reliability.

Summary

Add Code entry to self-driving docs menu

Adds a Code link to the self-driving docs products menu, enabling users to access the Code product directly.

Health Assessment

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

Summary

Sync ClickHouse versions from infra

Automated sync of ClickHouse versions and settings from infra, ensuring CI uses correct versions.

Health Assessment

• • Quick automated sync with minimal changes, fast review and merge.

Summary

Document pytest GC boot window plugin

This PR documents the pytest_boot_gc plugin that opens the GC boot window before django.setup() runs in test processes. It explains how to enable the plugin and highlights the ~0.2–0.4s per-process performance improvement.

Health Assessment

• • Quick documentation update with minimal review time, indicating low complexity and high confidence.

Summary

Raise wizard daily product cost cap

Increases wizard product cost cap to prevent gateway rejections, ensuring uninterrupted service.

Health Assessment

• • Fast turnaround with minimal changes; AI-assisted code change resolved issue quickly.

Summary

Reduce pytest boot cost by disabling plugins

Optimizes CI test performance by disabling unused pytest plugins and adjusting GC timing, cutting per-process boot time by ~6 seconds.

Health Assessment

• • Fast turnaround with minimal changes; single review cycle and no production code touched.

Summary

Support agent settings feature

Adds beta support for agent settings in the conversations UI, improving customization and user experience.

Health Assessment

• • The PR was merged within an hour of opening, with only two brief reviews and no significant rework, indicating a smooth, low-risk change.

Summary

Add pattern diff engine for log comparison

Enables comparison of log patterns across time windows, powering anomaly detection and incident analysis. This new engine supports UI compare mode and MCP tool integration.

Health Assessment

• • Rapid review and merge within 22 hours indicates high confidence and minimal friction. The PR introduced a large codebase change but required only a single review cycle, suggesting well‑structured implementation.