jobizzness/bench

AI coding that actually makes you move with superhuman speed.

Stars: 1
Forks: 0
Open beginner issues: 0
Indexed issues: 47
Avg merge: 1h 21m
Merged PRs (30d): 41

Dominant language: TypeScript
License: MIT
Last GitHub push: Sep 13, 2026
Latest indexed: Sep 18, 2026
Contributing guide: Contributing guide
Code of conduct: No code of conduct
Beginner labels: No beginner labels indexed

Open All

47 open issues indexed

Cost: the bill is 210:1 input to output — re-sending the conversation, not the work Open

jobizzness/bench#33 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
entry.turnsTaken goes stale for any specialist stopped and revived within one daemon uptime Open

jobizzness/bench#39 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Tabs crash with "Session ID … is already in use" citing their own freshly created id, after accepting a brief Open

jobizzness/bench#43 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Should the Firestore mirror be end-to-end encrypted? Open

jobizzness/bench#48 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Push notifications when a specialist is waiting on you Open

jobizzness/bench#49 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Turn on the Firebase project and prove remote identity against it for real Open

jobizzness/bench#52 · 6 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Prove the Firestore wire against a real phone and a second machine Open

jobizzness/bench#55 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Nothing tells you the cockpit on your phone is running an old build Open

jobizzness/bench#63 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
The phone screenshot harness gets rebuilt from scratch every time, and has a trap in it Open

jobizzness/bench#67 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Answers can still fail silently — #60 fixed two of four call sites Open

jobizzness/bench#69 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
A crashed tab says "turn failed: success", which tells you nothing Open

jobizzness/bench#70 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Four test files hard-code a shared /tmp path as the daemon home Open

jobizzness/bench#71 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
bench-cli tests time out under load — every case spawns a fresh `npx tsx` Open

jobizzness/bench#72 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Changing the Anthropic key in Settings does not change which account specialists spend Open

jobizzness/bench#73 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Two elements share the id "empty", and the phone's CSS hits the wrong one Open

jobizzness/bench#77 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
The phone's panes move: directional transitions and sheets that rise Open

jobizzness/bench#81 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
bench restart --build cannot build, so it never restarts Open

jobizzness/bench#85 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Checklist and queue skeletons — same tri-state gap as useThread had Open

jobizzness/bench#87 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
A phone can run a stale bundle forever, and nothing says which one it is on Open

jobizzness/bench#89 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Answering a decision from the composer throws unhandled on a dead link, instead of failing like every other send Open

jobizzness/bench#96 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
The picker shows Medium for a specialist running at the global effort Open

jobizzness/bench#101 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Stopping a tab before its first turn ends makes the next prompt crash with "Session ID … is already in use" Open

jobizzness/bench#105 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Extract a Session interface so registry.ts stops depending on ClaudeSession Open

jobizzness/bench#106 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
DevinSession: run a specialist on `devin acp` as a second runtime Open

jobizzness/bench#107 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Switching a specialist between Devin and Claude crashes it: "No conversation found with session ID" Open

bug

jobizzness/bench#113 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Devin has no model picker and no way to reach `adaptive` — every Devin tab silently runs on swe-2-high Open

bug

jobizzness/bench#114 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Devin turns report zero tokens and zero cost — the usage data is on the ACP wire and is being dropped Open

bug

jobizzness/bench#115 · 8 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
A stalled Devin turn hangs forever: no progress, no timeout, and inbound ACP requests are dropped Open

bug

jobizzness/bench#116 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Confirm the wire shape of Devin's turn_stats/billingInformation notifications, to price Devin turns Open

jobizzness/bench#117 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Two model-picker tests are red on `devin` — one stale count, one that may mean clicking Devin does not post Open

bug

jobizzness/bench#118 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Ditch the Gemini/OpenRouter proxy — Devin CLI covers other models Open

jobizzness/bench#121 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Daemon: emit structured file-edit events on /events, carrying the absolute path Open

jobizzness/bench#122 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
VS Code extension: open the file a specialist just edited Open

jobizzness/bench#123 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
README claims a specialist may not push a branch; the gate that did that was removed Open

jobizzness/bench#124 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Local tabs relay through Firestore instead of the daemon, and once they do it never unsticks Open

jobizzness/bench#126 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
VS Code: a Bench sidebar view listing each specialist on this project and its changed files Open

jobizzness/bench#128 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Cockpit: a button that points an open VS Code window at a project Open

jobizzness/bench#129 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
"Session ID … is already in use": the daemon heals the claim but never acts on it, so the tab stays crashed until a human prompts it Open

jobizzness/bench#130 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
DevinSession: a hung setup handshake (session/new / session/load) has no timeout Open

jobizzness/bench#131 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Devin's model picker is empty when `devin models list` fails — fall back to the ACP configOptions Bench already reads Open

bug

jobizzness/bench#133 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Let the developer pin one Anthropic key, and make the Profile dialog look like the rest of the app Open

jobizzness/bench#135 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Devin's model picker shows the slug twice: every family label falls back to its id Open

jobizzness/bench#136 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Devin ACP session ended by 60s stall watchdog during blocking get_output wait Open

bug

jobizzness/bench#143 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
docs/STATUS.md has no account of the Devin runtime Open

jobizzness/bench#144 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Roster: let the developer drag the left pane wider, and start it 50px wider than it is Open

jobizzness/bench#145 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
Update button refuses on untracked files, calling them "uncommitted changes" Open

jobizzness/bench#149 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026
A slow update reports failure in the cockpit while it is still succeeding Open

jobizzness/bench#150 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 18, 2026

A short digest of beginner-friendly GitHub issues.