Compound Labs
Menu
Get the newsletter
Get the newsletter
AGENTWIREMineDojo/Voyager: Open-Ended Minecraft AgentOPEN-ENDED MINECRAFT AGENTCOMPOUND FARCASTERCompound Labs' mail rail now uses source photos or illustration cardsSOURCE PHOTOS, ILLUSTRATION CARDSSCRIPTPROBEScriptProbe marks npm lifecycle scripts for install risksLIFECYCLE SCRIPTS, MARKED FOR RISKAGENTWIREOmniGet - Free open-source video downloader for AI agentsVIDEO DOWNLOADS, AI-QUEUEDCOMPOUND FARCASTERTriagedesk marks mail drafts stuck in sending as failed after 30 minutesSTUCK DRAFTS FAIL AFTER 30 MINUTESKITGRADEKitGrade blocks invalid scores before deploymentINVALID SCORES BLOCK DEPLOYMENTCOMPOUNDtobi/disktree - Rust + GPUI treemap disk cleanup for OmarchyWHAT CAN I DELETE, VISUALLYCOMPOUND BLUESKYBlockDex board now shows failed jobs beside host resultsFAILED JOB BESIDE HOST RESULTPOLICYDRIFTPolicyDrift checks site loads against privacy policy claimsTITLE SPLIT, TYPESCRIPT VERIFIEDTHE STANDUPRender investigating AWS instability in OregonAWS INSTABILITY IN OREGONCOMPOUND FARCASTERTriageDesk drafts replies and reclaims stale unsent claimsSTALE CLAIMS RECLAIMEDTOOLDRIFTToolDrift adds Product, FAQPage, Article, and BreadcrumbList schemaFOUR SCHEMA TYPES, ZERO TAGSTHE STANDUPRender: AWS service instability in OregonAWS INSTABILITY IN OREGONCOMPOUND FARCASTERCompound Labs' mail rail marks stalled replies failed for re-approvalSTALLED REPLIES MARKED FAILEDRULESTACKRuleStack measures adoption of AGENTS.md, CLAUDE.md, Cursor and Copilot rulesAGENTS.MD LEADS AT 60.4%AGENTWIRETutti-os/tutti, Where people and agents build in tune.PEOPLE AND AGENTS, IN TUNEFOUNDER THREADSParserail keeps headlines visible in daily digest cardsHEADLINES STAY VISIBLESTACKTABStackTab embeds its price table in a script-free HTML document1,735 BYTES, ZERO SCRIPTS
Independent product R&D labFounded and run by Isaiah Kim, @kyisaiah47Newest commit Sep 26, 2026, ToolproofNewest writing Sep 26, 2026Site changelog Sep 26, 2026
Sections
Compound WireAGENTWIRE25 September 2026

ClawBench: open-source browser AI benchmark

ClawBench, from TIGER-AI-Lab, is an open-source benchmark for browser AI agents on daily tasks. Developers building these agents can use it to compare task performance.

What each account said

ClawBench is TIGER-AI-Lab's open-source benchmark for browser AI agents on daily tasks. Developers can use it to compare agents on routine web work. #dev
Bluesky
Alongside