Cockpit · main
How long each merged pull request was written against how long it waited on the harness, how many rounds it took, and which check held each one. Read the box below before the totals: it says what they leave out.
One point per merged pull request read, not one window, positioned by when it merged — so a trend is something to see rather than a pair of snapshots to compare by eye. Each panel's own size lane carries the pull request's lines changed. Zoom in with the tabs below; every day's data is already on this page, so nothing is fetched to do it.
Minutes spent writing and fixing against minutes spent in the harness, the rounds plus local review. Every pull request pays a fixed harness cost, so a small change sits above the dashed line whatever it did; read the minutes beside the ratio. No line is drawn as healthy, since there is no accepted benchmark for this ratio: the reference is the window’s own median.
Median ratio 1.1× over 43 pull requests.
Scale to 2h 00m on both axes; 3 dots beyond it, drawn hollow on the edge. The three furthest above the dashed line are named. Hover or focus a dot for its figures; it opens the pull request.
66 merged pull requests in this window carry no session record and are not on the chart.
| Pull request | Coding and fixing | In the harness | Ratio | Rounds | |
|---|---|---|---|---|---|
| #607 Search the Move to / Add to picker, and tell its dashboards from its panels at a glance | 24m 53s | 31m 06s | 1.2× | 2 | |
| #606 Show which app captured an item | 11m 08s | 12m 12s | 1.1× | 1 | |
| #597 Wear a workspace's colour in the dashboard band, not across the page | 12m 05s | 41m 55s | 3.5× | 3 | |
| #596 Put the account's settings at the foot of the workspace's … menu, and drop the gear | 10m 44s | 15m 17s | 1.4× | 1 | |
| #593 Restyle the agents' dock to match the agreed POC | 1h 12m | 41m 17s | 0.57× | 3 | |
| #591 Send an item's attachments along when an agent starts | 1h 11m | 2h 49m | 2.4× | 2 | off the chart |
| #592 See on the item when Claude is waiting on you | 2h 09m | 34m 39s | 0.27× | 2 | off the chart |
| #590 Drop an agent on an item to start a Claude Code session on it | 1h 14m | 33m 16s | 0.44× | 2 | |
| #589 Enable prompt caching on the note-cleanup prompt, restructured so the fixed content is a stable prefix | 53m 54s | 1h 57m | 2.2× | 2 | |
| #585 Debounce the settle-triggered repropose fan-out across a real time window | 30m 59s | 1h 00m | 1.9× | 1 | |
| #586 Record every panel refresh in the rewrite history | 40m 50s | 41m 30s | 1.0× | 2 | |
| #587 Keep your agents in a dock, and choose which each dashboard shows | 1h 02m | 55m 02s | 0.88× | 2 | |
| #588 Ask a settled filing's refresh for the panel alone, on Haiku 4.5 | 41m 43s | 21m 21s | 0.51× | 1 | |
| #579 Connect a workspace to Claude Code | 1h 09m | 55m 32s | 0.80× | 3 | |
| #576 Give the open workspace and dashboard their own "…", and split the header's menu into settings and you | 57m 15s | 35m 56s | 0.63× | 2 | |
| #578 Mark an item In progress, and see since when | 56m 26s | 25m 42s | 0.46× | 1 | |
| #577 Hand a panel's scroll off to the dashboard once it runs out of room | 28m 55s | 26m 22s | 0.91× | 1 | |
| #564 Publish nightly's reports the same night, and make a manual run go live too | 6m 34s | 24m 34s | 3.7× | 1 | |
| #561 Replace the lead-time report's figures/harness/numbers tables with a single evolution chart | 13m 37s | 32m 32s | 2.4× | 1 | |
| #559 Drop files and paste images while capturing a message | 32m 44s | 36m 35s | 1.1× | 2 | |
| #558 Stop a tests/e2e/ change forcing the Test job's Vitest packages into full | 10m 18s | 14m 36s | 1.4× | 1 | |
| #555 Show E2E in the test selection report | 9m 00s | 1h 01m | 6.9× | 1 | |
| #554 Publish a test selection report | 43m 45s | 1h 18m | 1.8× | 2 | |
| #553 Say when the free tier's daily reads are spent, instead of blaming an update | 31s | 51m 01s | 100× | 1 | |
| #552 Stop every push-stream poll reading all of an account's note readings | 1m 52s | 25m 03s | 13× | 1 | |
| #547 Capture over the screen you are on, and open it with C | 1h 26m | 27m 43s | 0.32× | 1 | |
| #544 Record what CI's Test job selected, and why, on every run | 36m 53s | 11m 40s | 0.32× | 1 | |
| #546 Collapse the Inbox to a chip in the Dashboard bar, and open it again with one press | 20m 57s | 13m 52s | 0.66× | 1 | |
| #545 Flag the users nobody has signed in as for months, on the admin page (#344) | 15m 10s | 25m 55s | 1.7× | 1 | |
| #543 Keep the Capture button in reach on a phone | 6m 35s | 16m 40s | 2.5× | 2 | |
| #542 Offer Capture from the installed app's icon menu | 3m 24s | 11m 59s | 3.5× | 1 | |
| #538 Run only the affected specs in CI's E2E (F3) job on a pull request | 22m 23s | 12m 50s | 0.57× | 1 | |
| #533 Remove the two learning settings screens, and the commands that write to them | 28m 01s | 45m 42s | 1.6× | 1 | |
| #532 Embed an image inline in an item's description | 29m 14s | 18m 00s | 0.62× | 1 | |
| #531 Choose how a Filter's rows are sorted | 16m 10s | 12m 27s | 0.77× | 1 | |
| #530 Sort a panel of items by the fields you choose | 2h 14m | 16m 01s | 0.12× | 1 | off the chart |
| #529 Change an item's type, and its status, from its form, and see where it is shown | 1h 37m | 24m 25s | 0.25× | 2 | |
| #525 Scroll the dashboard or a panel while dragging near its edge | 13m 09s | 22m 06s | 1.7× | 2 | |
| #521 Plot time coding against time in the harness | 12m 57s | 14m 11s | 1.1× | 1 | |
| #519 Run the lead-time report nightly, and publish it beside the other two | 3m 24s | 17m 08s | 5.0× | 1 | |
| #518 Draw the lead-time page from the model | 13m 39s | 20m 19s | 1.5× | 1 | |
| #517 Work out each merged pull request's rounds and what held them, and answer with the model | 24m 20s | 14m 01s | 0.58× | 1 | |
| #516 Record what a session did, on the pull request it opens | 41s | 15m 16s | 22× | 1 |
Every strip is on one scale, up to 4h 00m. Time away is not on it. A round is split into the checks that held it, in the order they finished — hover a slice for which one and how long; open a row for its rounds in words.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
The session record's start is after the pull request's first commit, so it is ignored.
Clipped: 4h 00m on the scale in all (time away is not counted), the first 4h 00m shown. The strip ends at the edge of the scale, not at the merge.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
Clipped: 5h 00m on the scale in all (time away is not counted), the first 4h 00m shown. The strip ends at the edge of the scale, not at the merge.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
Clipped: 4h 25m on the scale in all (time away is not counted), the first 4h 00m shown. The strip ends at the edge of the scale, not at the merge.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
Clipped: 5h 39m on the scale in all (time away is not counted), the first 4h 00m shown. The strip ends at the edge of the scale, not at the merge.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
Clipped: 4h 46m on the scale in all (time away is not counted), the first 4h 00m shown. The strip ends at the edge of the scale, not at the merge.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
Clipped: 46h 26m on the scale in all (time away is not counted), the first 4h 00m shown. The strip ends at the edge of the scale, not at the merge.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.
No session record: the time starts at the first commit, so any coding before it and the local reviews are not recorded.