BenchFlow
RL Environments for Coding Agents
Real-world coding tasks from production TypeScript repositories. Train agents on actual engineering problems via PR Mirroring.
1.2M+
Combined Stars
30
Repositories
100%
From Real PRs
Validated Task Instances
Human-reviewed training tasks with verified fail-to-pass test coverage
Add Fixed Date Range with Time Picker
FEATUREImplement a new date filter option that allows users to select both date and time for analytics queries
GeoIP Properties Trigger Person Updates
BUG FIXFix person property updates when allowed GeoIP properties change, ensuring all related properties are updated together
Partner Rewards Commission Verification
FEATUREImplement E2E tests for partner lead and sale rewards with country-based commission modifiers
Team Member Permission-Based Access Control
FEATUREAdd granular permission controls for team member management with role-based UI visibility
Sanitize App Data for Viewer Security
BUG FIXRemove sensitive credential data from app responses before sending to client viewers
Calendar Subscription Service Feature Flags
ENHANCEMENTAdd feature flag support to calendar subscription service with repository pattern refactoring
Reactive Actions Dependency Detection
FEATUREDetect and warn when reactive actions have both .run and .data dependencies causing potential infinite loops
JS Function Run Behaviour Settings
FEATUREAdd ON_PAGE_UNLOAD run behaviour option for JS functions with feature flag control
Source Repositories
Production codebases with active development and test coverage
calcom/cal.com
39kOpen-source scheduling infrastructure. The Calendly alternative.
twentyhq/twenty
37kOpen-source CRM. Modern alternative to Salesforce.
makeplane/plane
40kOpen-source project tracking. JIRA/Linear alternative.
Infisical/infisical
24kOpen-source secret management platform for teams.
dubinc/dub
23kOpen-source link management. Bitly alternative with analytics.
activepieces/activepieces
19kOpen-source workflow automation. Zapier alternative.
documenso/documenso
12kOpen-source document signing. DocuSign alternative.
formbricks/formbricks
12kOpen-source survey platform. Qualtrics alternative.
midday-ai/midday
13kFinancial tools for freelancers. Invoicing and tracking.
openstatusHQ/openstatus
8kSynthetic monitoring and status pages. Open-source.
langgenius/dify
120kLLM app orchestration platform with RAG and agents.
lobehub/lobe-chat
69kAI Agent Workspace with multi-provider support and RAG.
FlowiseAI/Flowise
47kVisual AI agent builder with drag-and-drop interface.
mckaywrigley/chatbot-ui
33kAI chat interface for any model. ChatGPT-style UI.
excalidraw/excalidraw
112kVirtual whiteboard for hand-drawn diagrams.
toeverything/AFFiNE
60kKnowledge base with docs and canvas. Notion + Miro alternative.
tldraw/tldraw
44kInfinite canvas whiteboard SDK for developers.
steven-tey/novel
15kNotion-style WYSIWYG editor with AI autocompletion.
supabase/supabase
94kOpen-source Firebase alternative with Postgres.
hoppscotch/hoppscotch
77kAPI development ecosystem. Postman alternative.
strapi/strapi
71kLeading open-source headless CMS. REST + GraphQL.
payloadcms/payload
39kFullstack Next.js CMS framework.
refinedev/refine
34kReact framework for admin panels and internal tools.
unkeyed/unkey
5kAPI key management and rate limiting platform.
appsmithorg/appsmith
39kLow-code platform for internal tool building.
ToolJet/ToolJet
37kOpen-source internal tool builder with AI.
PostHog/posthog
30kProduct analytics, feature flags, and experimentation.
medusajs/medusa
31kComposable headless commerce platform.
vercel/commerce
14kHigh-performance Next.js e-commerce starter.
sadmann7/skateshop
6kNext.js 14 e-commerce with Server Actions demo.
How It Works
PR Mirroring creates realistic coding tasks from real engineering work