Repository navigation
MC-37: show all of an agent's running work, and put the link first in Running now - #8
Merged
Merged
Conversation
…kind, title and question key (MC-37) The shared query in agent_runs/running.py describes a run at work now (its kind, title, question key and link) and will also feed the Agents pages. RunningItem moves to agent_runs/schemas.py; the Dashboard keeps exposing it under the same name. A cron run now shows as a cron with a working link, in scope through its project, instead of a question linking to /questions/None. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…r and admins (MC-37) AgentOut.runs uses the shared running query. Other viewers get an empty list, so the project's agent list does not reveal work in projects they cannot see. A run in a project the viewer cannot open keeps its text but has no link. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…on link, on two lines (MC-37) Line 1 is the link and title, line 2 the agent, status and time. The agent's name is what shortens, so a long name no longer pushes the link out of view. The stage labels, the status text and the ticking clock move to lib/runs.ts, and RunLink draws what a run works on (a task's key, a chat's or cron's tag and title, a question's key, a draft as text), so the Agents pages can reuse them. The API types are regenerated for the new RunningItem fields. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…first and the rest behind +N more (MC-37) The two Agents pages need to show every run an agent is working on. WorkingOn draws that list from the shared RunLink and run status, with the running time. Collapsed, it shows the newest run and opens the rest in place behind a "+N more" button, so the Admin table keeps its rows one height. The open state lives in the component and is not saved. No run shows a dash. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ow many runs are in use (MC-37)
current_run is replaced by runs, which the shared query fills. The admin
endpoint no longer runs its own two queries for the newest task or question
run. The page's "Working on" column shows the newest run as a link, with the
rest behind "+N more", and the status cell says how many runs are in use
("3 of 3 running").
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…-37) The agent's card now has a "Working on" section that lists every run the agent is working on, each with its link and status. The section is hidden while the agent runs nothing. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The parallel spec already gets one agent running three chats at once. It now checks that the agent's card on My agents lists all three, each as a link, and that Admin → Agents shows the newest run and opens the other two behind "+2 more". Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
mmhy2003
commented
Nov 18, 2026
mmhy2003
left a comment
Owner
Author
There was a problem hiding this comment.
Verdict: approved
The change matches the approved design (mockup v4) and plan, step for step:
- Backend:
agent_runs/running.pyis now the one query for "a run at work now". It covers tasks, chat rounds, questions, crons and drafts, newest first. The Dashboard'srunning()uses it, and cron runs are now in scope through their project, which fixes the old/questions/Nonelink.AgentOut.runsis filled only for the agent's owner and for admins. A run in a project the viewer can't open keeps its text but loses its link.current_runand its two queries are gone. - Frontend:
lib/runs.ts,RunLinkandWorkingOnare shared. "Running now" rows have two lines with the link first, and the agent's name is what shortens. Admin → Agents shows the newest run with "+N more" / "Show less" (witharia-expanded) and "3 of 3 running" in the Status cell. My agents has a "Working on" section, hidden when the agent runs nothing. - Tests: the new backend tests cover the admin list, mine, and project visibility. Vitest covers RunLink, WorkingOn, RunningNow, the Admin page and AgentsTab. The
parallel.spec.tsbrowser test checks both Agents pages with three runs at once. - CI: every check passes (backend, frontend, runner ×3, release-tools).
I found nothing that blocks the merge.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
MC-37: show all of an agent's running work, and put the link first in "Running now"
What changed
app/agent_runs/running.py) now feeds both the dashboard and the agent lists.AgentOut.runsreplacesAdminAgentOut.current_run./questions/None.lib/runs.ts,components/runs/RunLink.tsxandcomponents/runs/WorkingOn.tsx.api-types.tswas regenerated.Why
The task asked to show all of an agent's concurrent work instead of only one item, and to put the link before the agent and status on the dashboard. Before this change, a long agent name pushed the link out of view.
Design and plan:
docs/superpowers/specs/MC-37-design.md,docs/superpowers/plans/MC-37-plan.md.How it was tested
npx tsc --noEmit,npx eslinton the changed files andnpm test(Vitest) all pass, 1262/1262.uv run pytest -n 8gave 2045 passed and 1 failed. The failure istests/dashboard/test_speed.py::test_dashboard_answers_under_300ms_with_10k_tasks, a 300 ms timing limit, which took about 0.40 s.aeedef6, before any MC-37 change (0.33–0.37 s), so the cause is the test machine (a freshly recovered dev database, high load), not this branch.parallel.spec.tswas updated to check that two runs at once both show on the Agents page) and the runner's go tests (the runner hasn't changed). Please run the e2e suite before merging if needed.🤖 Generated with Claude Code