Issues

• [Feat] Confidence-gated deferral: hand low-confidence browser steps to the chat model and measure the trade-off

  enhancement
  ThinkFlowLab/system1-agents#94 · Filter by author Birfy opened on Oct 22, 2026

• [Deps] Pin Jev on the direct TypeSafe path (it sends jev-latest) and re-run benchmarks on jev-1.14.0

  enhancement
  ThinkFlowLab/system1-agents#93 · Filter by author Birfy opened on Oct 22, 2026

• [Eval] Measure option-order sensitivity of each decision backend (flip rate under reversed/shuffled options)

  enhancement
  ThinkFlowLab/system1-agents#90 · Filter by author Birfy opened on Oct 19, 2026

• [Decisions] Make confidence mean the same thing on every backend (Laya reports normalized entropy, others max p or Jev's margin)

  enhancement
  ThinkFlowLab/system1-agents#89 · Filter by author Birfy opened on Oct 19, 2026

• [Feat] Add a served OneJev multimodal decision backend (System One + media)

  enhancement
  ThinkFlowLab/system1-agents#88 · Filter by author Birfy opened on Oct 17, 2026

• [Eval] Extend injection_guard measurement to BrowseSafe++ (matched controls + multi-view pages)

  enhancement
  ThinkFlowLab/system1-agents#87 · Filter by author Birfy opened on Oct 17, 2026

• [MCP] Keep long run_agent calls inside host tool timeouts: progress, bounded lock wait, cancellation, documented settings

  enhancement
  ThinkFlowLab/system1-agents#83 · Filter by author Birfy opened on Oct 15, 2026

• Security Notice: Potential Decision Hijacking Risk in Jev-Based Applications

  ThinkFlowLab/system1-agents#80 · Filter by author yuchenWang188 opened on Oct 14, 2026

• [Help wanted] Live multi-question game dashboard (Mario-style) on the decision slot

  enhancementhelp wantedhigh prioritythinkflowlab
  ThinkFlowLab/system1-agents#79 · Filter by author hsliuustc0106 opened on Oct 14, 2026
  1

• [Eval] Measure the injection guard on page-length inputs from BrowseSafe-Bench

  enhancement
  ThinkFlowLab/system1-agents#75 · Filter by author Birfy opened on Oct 14, 2026

• [Deps] Bound mcp below 2 and record what blocks the move to the 2026-07-28 SDK

  dependencies
  ThinkFlowLab/system1-agents#74 · Filter by author Birfy opened on Oct 14, 2026

• [Security] Define injection-guard handling for truncated and uncertain outputs

  ThinkFlowLab/system1-agents#57 · Filter by author Birfy opened on Oct 9, 2026

• [Tracking] Roadmap: applications, benchmarks, model backends and System1-Omni

  ThinkFlowLab/system1-agents#54 · Filter by author hsliuustc0106 opened on Oct 8, 2026
  1

• [Feat] Connect desktop screenshot decisions to System1-Omni Cua-S1 4B

  ThinkFlowLab/system1-agents#44 · Filter by author QianCyrus opened on Oct 6, 2026

• [RFC] Local Cua-S1 4B and screenshot targets for #14 and #15

  ThinkFlowLab/system1-agents#30 · Filter by author QianCyrus opened on Sep 27, 2026

• [RFC] macOS desktop input and save flows for #15

  ThinkFlowLab/system1-agents#26 · Filter by author QianCyrus opened on Sep 27, 2026
  2

• [Feat] Integrate and evaluate Julia-1

  ThinkFlowLab/system1-agents#25 · Filter by author Yunaik opened on Sep 27, 2026
  4

  

• [Feat] Integrate and evaluate CLM-v0.1-8B through clm-serve

  ThinkFlowLab/system1-agents#24 · Filter by author Yunaik opened on Sep 27, 2026
  1

  

• [Feat] Extend desktop execution to typing, keyboard input, and scrolling

  ThinkFlowLab/system1-agents#15 · Filter by author Yunaik opened on Sep 25, 2026
  1

• [Feat] Integrate and evaluate Cua-S1 4B 0.2

  ThinkFlowLab/system1-agents#14 · Filter by author Yunaik opened on Sep 25, 2026

• [Eval] Add repeatable browser and desktop task suites

  ThinkFlowLab/system1-agents#13 · Filter by author Yunaik opened on Sep 25, 2026
  1

• injection_guard: a labelled set from InjecAgent and AgentDojo

  good first issue
  ThinkFlowLab/system1-agents#3 · Filter by author Yunaik opened on Sep 22, 2026
  4