Pull Request Explorer

Summary

Tag Max insight queries as PostHog AI source

Ensures AI‑driven insight executions are correctly attributed in analytics, improving visibility and reporting accuracy.

Health Assessment

• • Quick, single‑commit fix with minimal code changes and no significant review friction.

Summary

Warn before navigating away from unsaved dashboard edits

Prevents accidental loss of unsaved dashboard layout changes, improving user experience and reducing frustration.

Health Assessment

• • Quick fix with minimal code changes, single review round, no new tests added, low risk to production

Summary

Fix hyperlink selection for long URLs

Ensures pasted URLs over selected text become hyperlinks regardless of length, improving user experience in the conversations support editor.

Health Assessment

• • Rapid review and merge with minimal changes; multiple merges from master indicate active integration but overall scope and risk remain low.

Summary

Fix execute-sql prompt column mismatch

Corrects an example in the MCP execute-sql prompt that incorrectly referenced a non-existent column, preventing query failures. Ensures accurate guidance for users and avoids runtime errors.

Health Assessment

• • Fast review and merge with minimal changes indicates low complexity and risk.

Summary

Fix data warehouse search to match managed sources by display label

Improves search functionality in data warehouse by matching managed sources by their display labels, making it easier for users to find what they need.

Health Assessment

• • The PR was reviewed and merged quickly, indicating a smooth development process. The changes are focused and well-tested, reducing the risk of introducing new bugs.

Summary

Add view settings menu to logs viewer toolbar

Users can now access wrap and timezone settings from a single toolbar button, streamlining configuration. This reduces UI clutter and improves user experience.

Health Assessment

• • Quick approval with minimal rework indicates a straightforward UI enhancement.

Summary

Add retry for inline source picker failures

Fixes infinite loading skeleton in the data‑warehouse source picker, reducing onboarding abandonment and improving user experience.

Health Assessment

• • Quick, single‑commit fix with fast review; minimal code changes and low risk to production.

Summary

Recover double-encoded source config strings

Fixes data‑warehouse import failures caused by double‑encoded configuration, improving reliability of data pipelines.

Health Assessment

• • Quick fix with minimal changes and fast review, indicating low complexity and high confidence in the patch.

Summary

Fix unparseable source config handling

Ensures the data import pipeline fails cleanly when a source configuration cannot be parsed, preventing retry loops and reducing error‑tracking noise.

Health Assessment

• • Quick fix with minimal changes, approved within 18 minutes, demonstrating an efficient review process.

Summary

Fix CDC slot recreation retry on transient drops

Ensures reliable data sync by retrying slot recreation after transient connection drops, preventing customer‑visible failures.

Health Assessment

• • Fast review and merge cycle (16h) with minimal rework; the change is small, scoped, and well‑tested, indicating low risk to production.

Summary

Add reasoning effort capture to LLM gateway events

Enables cost and quality analysis by recording caller‑specified reasoning effort in LLM generation events. This allows slicing LLM cost by effort tier for better insights.

Health Assessment

• • Fast review and merge with minimal rework; the PR was completed in under 7 hours and required no additional review rounds.

Summary

Add integration created analytics event

Adds server‑side analytics capture for integration creation, improving funnel visibility for onboarding.

Health Assessment

• • Rapid approval with minimal changes; AI‑assisted code generation; low complexity and risk.

Summary

Fix experiments recalculation latency

Reduces final activity latency from ~10s to milliseconds, improving UI responsiveness.

Health Assessment

• • Fast cycle time and minimal rework indicate a straightforward performance fix.

Summary

Add cumulative query progress to experiments recalculation

Provides accurate live progress for experiment recalculations, improving visibility for users and reducing confusion caused by zero‑filled progress bars.

Health Assessment

• • Fast cycle time (9.3h) and short review window (4.7h) indicate efficient review and minimal rework. The PR adds moderate code changes across backend and frontend, with comprehensive tests ensuring reliability.

Summary

Add load bias to Cymbal resolution routing

Improves load balancing by biasing routing away from overloaded resolver pods, enhancing reliability and performance.

Health Assessment

• • Quick approval cycle with minimal rework indicates high confidence in changes.

Summary

Deduplicate marketing cost charts via precomputed view

Fixes double-counting in marketing analytics cost charts by introducing a deduplicated view, ensuring accurate reporting and preventing inflated metrics. This improves data integrity for marketing teams and reduces reporting errors.

Health Assessment

• • Fast review and merge with minimal rework; AI assistance streamlined testing and code quality checks.

Summary

Add ticket nudges docs for Slack integration

This PR updates support documentation to include the new ticket nudges feature for Slack integration. It adds a new section and updates existing tables to reflect four ticket creation methods.

Health Assessment

• • Fast merge with no review needed, indicating a straightforward documentation update.

Summary

Add typed column model and migration for logs

Provides a unified column model for logs, enabling future server‑computed custom columns and a migration path for existing users, improving scalability and maintainability. This change lays groundwork for future UI enhancements without breaking current functionality.

Health Assessment

• • Rapid review and merge with minimal changes indicates high confidence and low risk. The PR was AI-assisted, adding a new capability that is dormant and does not affect current user experience.

Summary

Add command to unstick wedged external data jobs

Provides an ops tool to recover stuck data jobs, preventing pipeline stalls and ensuring timely data syncs.

Health Assessment

• • Fast review and merge (18h cycle) with AI-assisted code generation; high line count but limited file scope; minimal risk to production.

Summary

Add email bounce prevented metric to workflows

Provides visibility into pre‑send email bounce prevention, helping teams monitor deliverability and reduce billing errors.

Health Assessment

• • Fast review and merge with minimal changes indicates low complexity and risk.

Summary

Add auto_publish option for cloud task runs

Allows users to automatically publish cloud task runs, streamlining workflow and reducing manual intervention.

Health Assessment

• • Fast review cycle and minimal iterations indicate a well-scoped change; extensive tests and clear documentation reduce risk.

Summary

Add Traces product page

New product landing page for Traces, providing an overview of PostHog's distributed tracing capabilities.

Health Assessment

• • The PR had a relatively fast cycle time of 48.9 hours, indicating efficient review and merge processes.
• • The number of commits and review rounds suggests some rework was necessary, but the overall process was still smooth.

Summary

Centralize ingestion warning types registry

Standardizes warning classification across ingestion pipelines, reducing errors and improving data quality for downstream analytics.

Health Assessment

• • Fast cycle time (1.5h) and single approval indicate a well‑understood change with minimal friction; the large scope reflects a comprehensive refactor of warning handling logic.

Summary

Fix egress budget backpressure error tracking

Prevents benign egress budget exhaustion errors from polluting error tracking, improving signal quality and reducing noise for ops.

Health Assessment

• • Quick review and merge within a day, minimal rework, low risk

Summary

Improve Snowflake error messages during setup

Enhances user onboarding by providing clearer Snowflake connection error messages, reducing confusion and abandonment.

Health Assessment

• • Quick review and merge, minimal changes, low risk.

Summary

Fix clear error for encrypted data-import config

Prevents leaking ciphertext and provides a clear error when decryption fails, improving reliability for data imports. Ensures data import jobs fail gracefully without exposing sensitive tokens.

Health Assessment

• • Fast review and single commit indicate a straightforward bug fix with clear test coverage.

Summary

Tag Gemini calls with ai_product for session summary

Ensures AI-generated video session summaries are correctly attributed to the session_replay product, improving analytics accuracy.

Health Assessment

• • Rapid, single-commit fix with AI assistance; minimal code churn and quick review.

Summary

Fix data imports crash with mixed list/scalar columns

Prevents data import failures when dealing with mixed list/scalar columns, ensuring robustness in data processing.

Health Assessment

• • The fix was fully autonomous, and the PR had a fast review time, indicating a straightforward and well-tested solution.

Summary

Fix Shopify store ID trailing dot handling

Ensures Shopify store IDs with trailing dots are accepted, preventing user errors during source creation. This reduces support tickets and improves onboarding.

Health Assessment

• • Quick fix with minimal changes and immediate approval indicates low risk and high confidence in correctness.

Summary

Validate LinkedIn Ads account ID numeric at setup

Prevents users from encountering sync failures by validating LinkedIn Ads account ID during setup, improving onboarding experience and reducing support tickets.

Health Assessment

• • Quick review and merge within 23 hours, minimal rework, small scope, low risk.

Summary

Expose v2 ingestion warnings as MCP tool

Adds a new MCP tool that exposes the ingestion‑warnings v2 API to agents, enabling them to diagnose missing events and merge failures directly from the platform.

Health Assessment

• • Fast cycle time (2.7h) and single review round indicate a smooth process with minimal friction; the change is medium‑to‑large in scope but well‑contained and low risk.

Summary

Fix double-scaling survey rates in MCP stats view

Corrects inflated dismissal and response rates in the Surveys MCP stats app by removing an erroneous double-scaling of already-percentage values.

Health Assessment

• • The PR was reviewed and approved quickly, with no significant rework needed.

Summary

Add Group view to logs/patterns bar

Enables users to view grouped logs directly, improving data exploration and reducing confusion.

Health Assessment

• • Fast cycle time (1.4h) and immediate approval indicate a straightforward UI change with minimal risk.

Summary

Fix table header overflow in insights

Prevents long header titles from spilling into adjacent columns, improving readability of insight tables.

Health Assessment

• • Fast review cycle (0h to first review) and minimal rework indicate a straightforward, low‑risk bug fix. The change is small in scope and only touches frontend UI logic.

Summary

Fix replay snapshot attribution on idle rotation

Ensures replay recordings are correctly associated with the new session after idle rotation, eliminating leading gaps in playback and improving user experience.

Health Assessment

• • Quick resolution with minimal rework; single bug fix merged within 16 hours.

Summary

Add messaging framework to brand foundations handbook

Adds the new messaging framework to the brand foundations handbook, ensuring consistent positioning across website, marketing, and engineering.

Health Assessment

• • Fast cycle time and minimal changes indicate a low-risk documentation update.

Summary

Move hobby deploy smoke test comment into shared CI report

Consolidates CI comments into a single shared report, reducing PR comment noise and improving visibility of test results.

Health Assessment

• • Quick single‑commit refactor with a 1‑hour review and 1.7‑hour cycle time indicates low complexity and minimal risk.

Summary

Deprecate trial evals in documentation

Remove references to trial evals from docs, aligning content with product changes.

Health Assessment

• • Merged within 1.3 hours with minimal changes and no review, indicating a straightforward documentation update.

Summary

Move plan doc out of backend

The PR moves a documentation file from an invalid backend directory to the correct plan directory, fixing a linting error that blocked CI builds. This ensures that backend CI runs smoothly and reduces manual intervention.

Health Assessment

• • Quick fix with minimal changes, no risk to production, CI stability improved.

Summary

Update marketing team page

Fixes marketing team page by updating quarter and removing duplicate section to reflect current information

Health Assessment

• • The PR was merged quickly with minimal review time, indicating a straightforward and low-risk change

Summary

Update LLM pricing models

Automated update of LLM cost data to keep pricing accurate.

Health Assessment

• • Automated PR with minimal changes merged quickly.

Summary

Fix Kafka flush before persisting alert state

Ensures alert notifications are reliably sent to Kafka before state is persisted, preventing silent loss and improving system reliability for users.

Health Assessment

• • The PR was reviewed and merged within 26.5 hours with minimal rework, indicating a straightforward reliability fix that was well-tested and had low impact on the overall codebase.

Summary

Align Rust HogVM datetime parsing to Node precision

Ensures Rust HogVM timestamps match Node baseline by truncating to millisecond precision, eliminating sub‑millisecond discrepancies that caused data mismatches. This fixes a large mismatch bucket in EU ingestion, improving data consistency.

Health Assessment

• • The PR resolved a precision mismatch between Rust and Node HogVM with minimal code changes and a quick review turnaround.

Summary

Add AI reply feedback for reviewers

Enables reviewers to rate AI-generated replies, improving quality measurement and integration with AI triage pipeline. Feedback is stored locally and sent to internal analytics for continuous improvement.

Health Assessment

• • Fast cycle time and minimal review iterations indicate a smooth integration of the new feedback feature.

Summary

Add loading state to cost section during fetch

Improves user experience by showing a loading indicator while cost data loads, preventing confusion about broken setup.

Health Assessment

• • Quick fix with minimal changes, approved in under 30 minutes, indicating low complexity and high confidence.

Summary

Format weekly digest counts as compact numbers

Improves readability of error tracking digest emails by formatting large numbers into compact form, enhancing user experience and reducing confusion.

Health Assessment

• • Fast cycle time and minimal review indicates straightforward change with low risk.

Summary

Remove stale overflowEnabled from tests

This change removes outdated overflowEnabled references from session replay pipeline tests, preventing TypeScript build errors that blocked all Node.js CI jobs. It ensures tests compile with the updated configuration, restoring CI stability.

Health Assessment

• • Fast cycle time and minimal rework indicate a straightforward bugfix with quick resolution.

Summary

Update size-gate limits based on denial outcomes

Adjusts PR size gate thresholds and exempts test files to reduce friction while maintaining risk controls.

Health Assessment

• • Rapid review and merge with minimal iterations, indicating a straightforward policy update with low risk.

Summary

Debounce sandbox composer to eliminate typing lag

Improves user experience in the PostHog AI sidebar by debouncing draft input, reducing lag and preventing race conditions.

Health Assessment

• • Fast review and merge (1.1h cycle time) indicates low complexity and high confidence in the fix.

Summary

Add claim-path flag for queue state reading

Enables efficient queue processing by reading state from denormalized columns, reducing load and improving performance while allowing safe rollback.

Health Assessment

• • Fast cycle time (3.4h) and minimal review friction indicate a well‑tested, low‑risk feature; extensive automated tests provide confidence in correctness.