All pull requests

Pull requests

• fix(desktop): stop the provider wizard crashing when left mid-step

  #1179·sosidudku1 opened Nov 11, 2026Collaborator·

  ·

• fix(cli): do not re-run a handoff when the TUI restarts after an update

  #1178·sosidudku1 opened Nov 11, 2026Collaborator·

  ·

• fix(tasks): end an approval wait when the task's time limit passes

  #1176·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• fix(overnight): one money cap for all /goal rounds, task total in the report

  #1175·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• fix(overnight): say when the Telegram report could not be sent

  #1174·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• fix(budget): skip the session-naming call after a budget stop

  #1173·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• fix(overnight): keep a task to the approval level it started under

  #1172·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• fix(overnight): say stopped and exit 2 when a limit ended the run

  #1171·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• fix(budget): count context compaction in the task spend

  #1170·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• Settings: add LM Studio, Ollama and Atomic Chat as local apps, never as cloud

  #1169·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• Desktop setup: Ollama, LM Studio and Atomic Chat move under Local models, plus use a .gguf you already have

  #1168·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• fix(desktop): label a reopened chat's steps the way they read live

  #1167·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• fix(desktop): swap in the polished reply live

  #1166·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• feat(desktop): regenerate the last reply by taking the turn back first

  #1165·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• fix(tui): make try again on the last question a real retry

  #1164·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• feat(http): label tool calls in a chat's history like the live stream

  #1163·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• feat(http): take the last plain question and answer back

  #1162·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• Studio: download a local model from Studio

  #1161·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• Studio: list only the models where the chat runs

  #1160·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• Studio: pick the image and video model where the chat runs

  #1159·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• Studio: notice a model that Generate just loaded

  #1158·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• Studio: say "this computer" instead of "this Mac" on Windows and Linux

  #1157·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• Bound compaction reasoning on Gemini and AI/ML API thinking models

  #1156·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• Reserve reply room for the reasoning budget on small windows

  #1155·sosidudku1 opened Nov 10, 2026Collaborator·

  ·

• Resolve qwen-think from a strict ChatML think template under a neutral alias

  #1154·sosidudku1 opened Nov 10, 2026Collaborator·

  ·