Pull Request Explorer

Summary

Add CLI subcommands for job rerun and restart

Enables users to re-run or restart completed jobs directly from the command line, improving automation and debugging.

Health Assessment

• • Fast cycle time and single review round indicate smooth integration.

Summary

Release version 1.706.1

This PR bumps the release to version 1.706.1, updating dependencies and configuration files to deliver the latest bug fixes and improvements to users.

Health Assessment

• • Fast cycle time and minimal review indicate a smooth release process.
• • AI review helped catch issues quickly.

Summary

Fix fork compare visibility for non-admins and stale-token superadmins

Ensures fork visibility correctly respects user permissions and stale superadmin tokens, preventing hidden changes and incorrect warnings.

Health Assessment

• • Rapid iteration with AI-assisted review, quick fix, but large code changes and multiple commits indicate significant refactoring of visibility logic.

Summary

Refactor sync git-deploy to separate commit/push

Fixes GPG signing failures by moving commit/push to caller, ensuring reliable deployments.

Health Assessment

• • Merged within minutes with minimal changes, indicating low complexity and high confidence in the refactor.

Summary

Fix GPG signing issue in git-sync

Resolves a GPG signing issue by reverting to a previous version of the git-sync script and updating the CLI to ensure reliable signing.

Health Assessment

• • The PR fixes a critical issue with GPG signing, and the solution involves reverting to a previous version of the script. The changes are relatively small, but the impact is significant.

Summary

Reduce slim image vulnerability surface

Reduces the dependency surface of slim Windmill images, making them more secure and reproducible.

Health Assessment

• • The PR reduces the vulnerability surface of the slim images and improves reproducibility, which is a positive change for the project's security and maintainability.

Summary

Gate Unix Symlink Test for Windows Builds

Fixes Windows build failures by gating a Unix‑specific symlink test, ensuring CI passes and improving reliability for Windows developers.

Health Assessment

• • Quick patch resolved Windows build issue with minimal code change, no rework required.

Summary

Release version 1.706.0

Publishes new release with updated dependencies and features.

Health Assessment

• • Release PR merged quickly with no review issues, indicating a streamlined release process.

Summary

Add optional disk-backed /tmp for nsjail

Enables disk-backed /tmp in nsjail sandboxes, reducing worker RAM usage and allowing disk-based temp storage, improving resource management for environments with limited RAM.

Health Assessment

• • Fast review and merge (7.6h cycle) despite large scope; minimal back‑and‑forth indicates clear implementation and testing. The change touches many files and adds significant logic, raising risk but mitigated by thorough unit tests.

Summary

Add UserDraft listing primitives

Enables global mode to discover, compare, and clear user drafts across workspaces, improving editor state consistency and reducing stale data.

Health Assessment

• • Fast cycle time with minimal review rounds indicates a straightforward, low-risk change.

Summary

Enforce RLS and scope checks for X-Resource-Path

Fixes an authorization bypass in the AI proxy by ensuring only users with read access to a resource can use its credentials, preventing privilege escalation.

Health Assessment

• • Quick fix with minimal rework, fast review, low risk.

Summary

Add non-root user support to debugger image

Enables the debugger container to run under non-root security contexts, improving Kubernetes compatibility and reducing permission errors.

Health Assessment

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

Summary

Add yolo mode for AI chat tools

Users can now toggle auto‑accept confirmations in AI chat, reducing friction and speeding up interactions.

Health Assessment

• • The PR was reviewed quickly with no issues, indicating a smooth process.

Summary

Add wmill init prompts and custom override slot

Splits AGENTS.md into managed and user-owned halves, adds CI-aware deploy guidance, and a dedicated wmill refresh prompts command

Health Assessment

• • The PR has a large scope with many files changed, indicating significant changes to the codebase. The review process was interactive with multiple comments and revisions, suggesting some friction in the review process.

Summary

Fix admin notification for wrong pod ingress

Admins now receive clear error messages when search traffic is routed to pods lacking the indexer lock, reducing troubleshooting time and preventing false health signals. Additionally, the index writer memory budget field now displays the correct default and unit, improving configuration clarity.

Health Assessment

• • Quick resolution with minimal rework; single review and no changes after approval.

Summary

Add UV_PYTHON_INSTALL_MIRROR env and setting

Enables operators to configure a private mirror for uv Python installs, improving reliability in air‑gapped or proxied environments.

Health Assessment

• • Fast cycle time and minimal rework indicate a smooth, low‑risk change.

Summary

Add CLI datatable serve and psql commands

Enables users to serve data tables via a Postgres‑wire proxy and launch psql directly, improving data access and developer workflow.

Health Assessment

• • Rapid iteration with AI‑assisted review led to quick resolution and merge within an hour.

Summary

Add Playwright MCP for frontend verification

Enables agents to drive a real browser for end‑to‑end frontend verification, improving QA automation.

Health Assessment

• • Quick merge with minimal changes and no review required, indicating low complexity and high confidence.

Summary

Release version 1.705.0

Deploys new release with updated features and bug fixes, updating dependencies and client libraries.

Health Assessment

• • Release was automated and merged quickly with minimal changes, indicating a smooth process.

Summary

Fix flow logs: flag noLogs jobs and lazy load logs

Adds an explicit sentinel for skipped logs, enabling the log panel to show a loader and lazily fetch full logs, eliminating empty log displays during flow testing.

Health Assessment

• • The PR had a long cycle time (~9 days) and slow review, largely due to multiple merge commits and a single review comment. Despite the moderate code changes, the extended timeline suggests coordination overhead rather than technical complexity.

Summary

Fix early return to respect failure_module result

Ensures sync endpoints return the failure handler’s output when an early-return node fails, improving reliability and correct error handling for users.

Health Assessment

• • Fast cycle time with a single review and minimal rework indicates a straightforward bug fix with low risk.

Summary

Refactor Google AI proxy handling

Moves Google AI workspace proxy handling from windmill-api to windmill-ai for better organization and maintainability.

Health Assessment

• • The PR was reviewed and merged within 3 hours, indicating a relatively smooth review process.
• • The refactor plan and test plan are well-documented, showing a clear understanding of the changes and their impact.

Summary

Add persistent in-editor drafts via UserDraft

Provides reliable autosave for all editors, preventing data loss on refresh or tab close, improving user productivity.

Health Assessment

• • Long review and merge times indicate significant rework and complexity; many commits suggest iterative development; large scope and high friction raise risk.

Summary

Collapse AI tool details automatically

Automatically collapse successful AI tool call panels to reduce clutter, while keeping test tools expanded for debugging.

Health Assessment

• • Rapid review and merge within 1.1 hours, minimal changes, no blockers.

Summary

Fix cgroup-aware DuckDB memory limit and allocator release

Ensures worker processes stay within container memory limits, preventing OOM kills and maintaining query spill behavior, thereby improving system stability and resource efficiency for users.

Health Assessment

• • The PR required several iterations and AI assistance to resolve complex memory management issues, yet it was merged within 22 hours, reflecting a relatively efficient resolution process.

Summary

Set cursor color in light editor theme

Improves editor usability by ensuring cursor visibility in light theme.

Health Assessment

• • Quick review and merge after a single commit; minimal changes.

Summary

Secure workspace envs, S3 handlers, job progress

This PR hardens workspace environment variable handling, adds preview-mode guards to S3 app handlers, and scopes job progress updates to prevent cross-workspace tampering, addressing multiple low‑severity vulnerabilities from the Q1 2026 report.

Health Assessment

• • Rapid response to vulnerability findings with minimal iterations and quick merge indicates an efficient review process.

Summary

Fix XSS via S3 download headers

Removes a stored‑XSS vulnerability that could let an attacker elevate privileges by exploiting download response headers, protecting admin sessions and preventing malicious content execution.

Health Assessment

• • The PR was merged within 18 minutes of opening, indicating a straightforward, low‑risk fix with no review or rework cycles.

Summary

Configure nsjail tmpfs size via instance setting

Allows operators to control memory allocation for sandboxed jobs, improving resource usage and consistency across languages.

Health Assessment

• • Rapid review and merge with minimal iterations indicates a straightforward change with low risk to production.

Summary

Add ghost breakpoint tooltip on gutter hover

Enhances debugging UX by showing a ghost breakpoint and tooltip, guiding users to set breakpoints easily.

Health Assessment

• • Multiple commits and a revert indicate iterative refinement; the 167‑hour cycle time suggests blockers or complexity in the debugging UI.

Summary

Health Assessment

Summary

Add waiting indicator and scroll polish to chat

Improves user experience by replacing misleading typing dots with a clear waiting indicator and polishing scroll behavior, enhancing clarity during tool confirmation and user prompts.

Health Assessment

• • Fast cycle time of 5.7 hours and only one review comment indicate a smooth review process with minimal friction.

Summary

Sync home search bar state to URL

Allows users to share and bookmark search results via URL, improving navigation and collaboration.

Health Assessment

• • Rapid AI-assisted change with minimal code churn and no review needed, leading to a fast merge.

Summary

Add CLI datatable and ducklake list/run commands

Adds new CLI commands to list and run SQL queries against datatables and ducklake, enabling users to manage data directly from the command line. This improves developer productivity by allowing quick data inspection and execution without leaving the terminal.

Health Assessment

• • The PR was merged quickly with minimal review, indicating straightforward implementation and clear documentation.

Summary

Fix SAML deep link preservation across round-trip

Ensures users return to intended deep link after SAML login, preventing navigation hang and security issues.

Health Assessment

• • Fast turnaround with minimal rework; AI review identified a single issue that was resolved quickly, indicating a clean, focused fix.

Summary

Resolve relative imports from local content in script/flow preview

Enables local edits to be used in script, flow, and app dev previews instead of last deployed versions

Health Assessment

• • The PR has a large scope with 13 files changed, but the review process was relatively smooth with a short cycle time of 28.9 hours

Summary

Add AI chat path linking to drawers

Enables AI chat messages to link workspace paths and open related items via drawers, improving navigation and user experience.

Health Assessment

• • The PR required multiple iterations after a single review, indicating significant rework, yet the overall cycle time remained within normal bounds.

Summary

Fix flow progress bar early‑stop and error handling

Ensures the flow progress bar accurately reflects completion and error states, improving user confidence in workflow execution.

Health Assessment

• • Quick resolution of two UI bugs with minimal code changes, indicating efficient review and low risk.

Summary

Fix autoscaling worker count logic

Accurate autoscaling of worker groups ensures optimal resource usage and prevents under‑scaling of custom worker groups.

Health Assessment

• • Quick fix with minimal changes, merged within minutes, indicating low complexity and high confidence.

Summary

Add oneshot system prompt with PR readiness guidance

Provides explicit guidance for webmux oneshot runs, ensuring PRs are only marked ready when confident, reducing review friction.

Health Assessment

• • Quick review and merge with minimal changes, indicating smooth process.

Summary

Visual redesign of AI chat interface

Enhances user experience by refreshing the AI chat UI, adding a streaming indicator, and improving scroll behavior, reducing confusion and improving usability.

Health Assessment

• • Fast cycle time and minimal review comments indicate a smooth, low‑friction process with a single commit and quick approval.

Summary

Derive Snowflake public key from private key

Users can configure Snowflake connections with only a private key, simplifying setup and reducing friction.

Health Assessment

• • Rapid review and merge indicate a straightforward change with minimal risk.

Summary

Harden main-branch guard and gate MCP tools

Strengthens repository protection by enforcing stricter git push rules and gating sensitive Claude AI connectors, reducing accidental force pushes and unauthorized tool usage.

Health Assessment

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

Summary

Add optional KV secret path prefix setting

Enables operators to keep existing Vault layouts and scope policies more tightly, improving security and flexibility for secret storage.

Health Assessment

• • Fast review and minimal rework led to a quick 9‑hour cycle time, indicating a smooth integration of the new feature.

Summary

Update webmux config for oneshot

Enable auto-cleanup and auto-worktree creation in `.webmux.yaml` for oneshot workflows, removing GitHub branches on merge and auto-creating worktrees for Linear issues for teams WIN and GIT.

Health Assessment

• • Rapid review and merge with minimal changes; AI review provided quick validation.

Summary

Enforce auth guards on app component preview execution

Secures app preview execution by enforcing proper authorization, preventing operators from running arbitrary code and ensuring token scope restrictions.

Health Assessment

• • Fast cycle time and minimal review rounds indicate efficient resolution of a critical security bug.

Summary

Add protection-rules pull/push CLI commands

Adds CLI commands to pull and push workspace protection rules, enabling version‑controlled rule management and improved CI integration.

Health Assessment

• • Rapid approval with minimal iterations indicates high confidence in the change.

Summary

Provision migrated DB for mutual-resource recursion test

Fixes CI panics by provisioning a migrated test database, ensuring the test runs against a proper schema.

Health Assessment

• • AI-assisted change resolved a CI failure with minimal code modifications, demonstrating efficient use of automated tooling.

Summary

Add observability for unavailable search index

Provides clearer error messages and a status panel for the indexer, enabling admins to quickly distinguish configuration issues from bugs and improving system reliability.

Health Assessment

• • The PR was reviewed and merged within 3.4 hours, indicating a smooth process with minimal rework. AI-assisted code generation likely contributed to the rapid turnaround.

Summary

Fix recursion depth limit for resource interpolation

Prevents API crashes from circular resource references, ensuring system stability and protecting against DoS attacks.

Health Assessment

• • Rapid AI-assisted fix with minimal code changes and no review required, resulting in a fast merge and low risk.