Skip to content

feat: add the Session Environments UI and dashboard API - #22238

Merged
kyleseaman merged 2 commits into
mainfrom
feat/session-environments-coder-ui
Oct 27, 2026
Merged

kyleseaman merged 2 commits into
mainfrom
feat/session-environments-coder-ui

Conversation

@kyleseaman

@kyleseaman kyleseaman commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: give the owner the controls for remote session environments: pick a Coder workspace for a chat, see and edit its files, open its terminal, and configure Coder profiles in Settings.

Part A (#19833) runs a bound session on its Coder workspace but has no way to bind one. This PR is the dashboard half. Part A is merged to main (ab2f15c57b), so this PR now targets main directly.

Behind the developer flag (off by default). We'll be testing this live on a real Coder deployment before it's turned on.

Why it matters

Without these controls the backend in #19833 cannot be used: nothing binds a chat to a workspace, and a remote chat's Files and terminal would act on the gateway.

Not a goal

What changed (motivation → approach → change)

Part What it does
src/kiro_crew/dashboard/handlers/session_environments.py, src/kiro_crew/dashboard/routes/sessions.py per-chat environment, browse, file save, readiness check, template list and workspace list routes, owner-only; a switch waits out a first start in progress
src/kiro_crew/dashboard/chat_api/slot_create.py a new empty owner chat starts on the default profile, unprewarmed
src/kiro_crew/coder/session_browse.py, src/kiro_crew/remote_session/workspace_tools.py, src/kiro_crew/session_environment.py, src/kiro_crew/config, src/kiro_crew/providers/acp_runtime_start.py, src/kiro_crew/coder/session_launch.py, config-baseline.json moved here from #19833, which has no caller for them: workspace browse, read and save (the save checks its folder), the terminal's coder ssh argv, the start-stage recorder, and the default_profile setting. Part A's test_the_guest_has_no_write_mode recorded that A had no caller for a guest write; this PR is that caller, so the test gives way to the write-path tests. A guest older than these files (one that answers "op must be …") is reinstalled once and the request retried
src/kiro_crew/dashboard/chat_turn/eager_spawn.py no prewarm for a chat bound to a profile, whatever armed it
src/kiro_crew/dashboard/handlers/terminal.py a remote chat's terminal is a shell in its workspace over coder ssh; never a local fallback (see below); no gateway completion or cwd
src/kiro_crew/dashboard/handlers/core_config_schema.py, src/kiro_crew/dashboard/handlers/core_config_patch.py, src/kiro_crew/dashboard/handlers/core.py the config editor validates session_environments with the loader's own normalizers; the success audit line for one of these structured writes records the path only, not the whole record; core.py's facade manifest names the merge helper it forwards
src/kiro_crew/dashboard/change_cards_settings.py a change card for one of those keys is checked by the same normalizer up front, so a value the route would refuse is never offered for approval
src/kiro_crew/change_card_catalog_kinds.py classes the four new editable session_environments keys for change cards: the URL and whole section as code-exec settings (they pick the machine that runs a session's commands), the default profile and shutdown switch as cannot-widen. These four entries sit in change_card_catalog_kinds.py because main's #22845 split moved both lists there from change_card_catalog.py; test_session_environment_paths_keep_their_class fails if one is dropped or reclassed
src/kiro_crew/docs, docs/feature-map/README.md generated UI index and settings registry; feature map rows
website/src/pages, website/src/components, website/src/api, website/src/hooks, website/src/utils, website/src/assets Settings card, composer chip and picker, start steps, remote Files panel, side-panel placement guard, terminal routing
website/src/i18n, website/src/test, website/eslint.i18n.config.js, website/scripts strings in all 12 catalogs, tests, settings extraction

Developer flag. With KIROCREW_DEV_MODE unset the gateway offers no profile, refuses a chosen one as unknown, and never binds a new chat (test_flag_off_offers_no_profile_and_binds_no_new_chat). The placement read skips the choices store, so an unreadable store can't lock a local chat's tabs (test_flag_off_placement_read_skips_an_unreadable_choices_store). The Settings tab also sits behind a new per-device Feature Previews switch, mc-preview-session-environments.

The gateway decides where a chat's terminal runs. A side-panel terminal names its chat (?slot=). With the flag on, the gateway refuses a local shell for a chat bound to a profile, even if the client didn't ask for the workspace shell (?remote_slot=). A bound chat gets a terminal_wrong_machine frame and the tab parks. A placement the gateway couldn't read (Coder or the choices store failed) also opens no local shell, but gets terminal_placement_unavailable, which the tab redials, so a brief outage never strands a live terminal. The bottom terminal names no chat and is the gateway's own, as on main.

File, diff and Git routes refuse a remote chat. /api/file-read, -raw, -stream, -watch, -write, -diff, /api/project/git, /git/status and /git/log read this machine's files, so a request whose ?slot= names a chat on a session environment is refused before the route runs (409 chat_runs_remotely, one fixed message that names no path; an unreadable placement counts as remote). The side panel's open, diff and save send their chat's slot=. Limits: a request without ?slot= passes unchanged, since the Files page, library previews and image thumbnails share these routes and carry no chat; a slot the gateway does not hold passes too; the routes stay owner-only, and a non-owner gets the route's own answer, so the guard never says where a chat runs. With the flag off nothing is refused and the placement store is not read. The client still keeps those tabs off a remote chat (a skeleton or error while placement loads or fails, then a pointer to the Files tab), and a terminal opened here before the chat was bound stops rendering. Code: dashboard/remote_chat_guard.py, wired in dashboard/routes/taskrunner.py.

Errors and labels. A failed switch shows inline beside the chip. Removing a profile says, beside Save, how many live chats it would strand. A failed check says what to do, with raw output under Details. A Remote Files conflict offers "Load workspace version (drops my edits)" and "Copy my edits", only on the failed file; a save made under a profile's old working folder is refused. A refused clipboard says so. The template field lists the deployment's templates, takes a typed name, and explains a failed load. "Ask the agent how to fix this" is left off only where an unsaved draft would be lost. Archived rows have "Reopen chat" and say how to keep the workspace. Over unsaved edits the buttons stay, disabled, and "Save or discard your changes first." is said once above the list. A refused Coder URL's reason sits under the box; the save banner just points there. A refused terminal offers no Reconnect. ! commands and Run in terminal refuse a chat on a profile. Disabled controls say why. One name per concept: profile (a template plus a working folder), the "Run this chat on" menu, a workspace per chat. The per-profile stop switch says it overrides the global one, with its own MicroVM line.

Backwards compatibility

Compatible: every new route is owner-only and answers "this machine" while the flag is off. A terminal, file, diff or Git request without ?slot= behaves as on main; ?slot= only adds refusals, and only while the flag is on.

Removes nothing: this PR only adds routes, query parameters and UI; nothing main accepts is rejected, renamed or removed.

Tests

  • Backend: test_session_environment_handlers (including the terminal refusal at the socket and helper level, each refusal's code, and the flag-off placement read), test_session_environments_config_api, test_default_session_environment, test_json_object_body_guard, test_remote_chat_file_guard (each guarded route refuses a remote chat; local, unknown and unnamed chats, a non-owner and the flag off reach the route; the refusal names no path).
  • Frontend: the card (URL field mark, Reopen chat and its failure, disabled-control reasons), chip (inline switch error), Files panel (copy failure, no hand-off on a stopped workspace), footer, side-panel placement (no-chat panel, the chat slot on each terminal tab), side-panel file requests carry their chat (usePanelDocumentActions.chatSlot), workspace rows name their chat and profile, plain Coder errors with the raw text under Details, terminal registry (slot / remote_slot query, terminal_wrong_machine parks the tab, terminal_placement_unavailable redials), CLI panel banner, settings coverage, i18n style.
  • Not run locally: full CI suites, native Windows/macOS.

Inherited CI reds (not this PR's)

These fail in files this PR does not touch. Each is re-run once, and none is fixed here.

  • Three strict XPASS tests in test/test_crew_log_fullscan_guard.py, which also fail on a clean main checkout (2356a1ba1c). They failed on Backend 3.12 shard 2 and Windows shard 2 at fa32704f0a:
    • test_a_cold_prime_copies_the_fold_state_a_fixed_number_of_times
    • test_a_ws_connect_does_not_fold_every_open_slot
    • test_a_ws_reconnect_with_nothing_new_reads_a_fixed_number_of_entries
  • macOS backend shard 3: every failure is "a lease outlived its test", cascading from leases left by tests in that same file. The counts were 286 at c399f47437 (after test_an_eager_advance_on_an_un…), 116 at fa32704f0a (after test_a_dispatch_view_for_one_unit_reads_a_fixed_number_of_units) and 287 at 2ac56e2150. The open PR test(crew-log): release full-scan guard handles at the test boundary #24590 releases those leases.
  • Windows backend shard 6: six test/test_slack_team_id_env.py::TestSpawnEnv tests fail with FileNotFoundError: [WinError 2]. That test file is new from main's feat: pass the slack workspace id to agents as KIROCREW_SLACK_TEAM_ID #22380, and main's own CI fails the same way on a3ee04f574 (job 121866397755) and 5285bc2878 (job 121891311949). This PR touches neither the test nor acp/client_start.py.
  • At 2ac56e2150 only:
    • test_a_first_history_page_in_a_new_process_reads_the_page_not_the_file, also a strict XPASS in that file. Main's test(crew-log): drop stale xfail on the first-history-page guard #24467 removed its stale xfail, and the current base includes that fix.
    • CodeQL Analyze (javascript-typescript): the analysis finished, and the job failed while uploading its results.

Manual verification

Kyle (PR owner) has run the Coder environment end to end against a live Coder deployment.

The two items once listed for before the flag is turned on are both handled in this PR: the live run above, and the gateway-side refusal of the file, diff and Git routes for a remote chat (see "File, diff and Git routes refuse a remote chat").

Screenshots / video

Rendered from this PR's HEAD in a Vite dev server on 127.0.0.1, with every /api/** call answered by a fixture (no Coder deployment). The whole-page shot is the built SPA behind the repo's loopback static server, with the same fixtures.

Settings › Session Environments, the whole page with its title:

Settings › Session Environments page, dark

The card: filled (archived workspace rows with "Reopen chat"), check failed, empty (the disabled check and switches say why), one profile open (the Working folder hint with its example, and the stop switch with its override hint), its template picker open and its list unavailable, a profile just removed (the live chats it strands, beside Save), Setup instructions, saved, save error (plain wording under the URL box with the gateway's own words under Details, a short banner, and the one "save first" line above the workspace list). Each workspace row names its profile and, for a live chat, the chat's title.

Session Environments card, dark
Workspace rows with their profile and chat, dark
Session Environments card, check failed, dark
Session Environments card, empty, dark
Profile editor with the persistence switch, dark
Profile editor, template picker open, dark
Profile editor, template list unavailable, dark
A profile just removed, dark
Setup instructions, dark
Configuration saved, dark
Save error, dark

A profile removed while the workspace list cannot be read (dark and light). The Manage section shows its error notice, and instead of claiming no chats are affected, the card says it could not check which chats still use the profile.

Removed profile, workspace list unreadable, dark
Removed profile, workspace list unreadable, light

Composer "Run this chat on" menu (template under each profile, selected one checked), the chip once fixed and while a reply runs, start steps, and the Feature Previews switch:

Environment menu, dark
Chip after the first message
Chip while a reply runs

While a reply runs the chip is disabled; its hover text (a native tooltip, which the browser does not draw into a capture) reads "Stop the current response to switch profiles". A failed switch names where it was going and where the chat still runs, beside the chip, wrapping instead of running past the composer row:

Chip after a failed switch
Start steps in the chat footer
Feature Previews switch

Remote Files: list, a file open, a save conflict with "Load workspace version (drops my edits)" and "Copy my edits", workspace stopped.

Remote Files list
Remote Files with a file open
Remote Files, save conflict
Remote Files, workspace stopped

A file with unsaved edits that a refresh now reads as past the 256 KiB limit (dark and light), or as binary: the editor and Save are gone, and the header offers "Copy my edits" beside "Discard my edits", and a line under it says why the file can no longer be edited here.

Remote Files, unsaved edits on a truncated read, dark
Remote Files, unsaved edits on a truncated read, light
Remote Files, unsaved edits on a binary read, dark

A local chat's side panel while placement is read, and after the read fails:

Side panel loading, local chat
Side panel error, local chat

A remote chat's side panel: an open file tab (it reads this machine's copy) shows the pointer to the Files tab instead. Then the workspace terminal opened from the Files panel's Terminal button (a scripted coder ssh socket stands in for the workspace), and a terminal opened on this machine before the chat was bound, which is no longer typed into once the chat is remote:

Remote chat, file tab withheld
Terminal in the chat's workspace
Gateway terminal withheld in a remote chat

Light theme:

Session Environments card, light
Check failed, light
Empty, light
Environment menu, light

Related Issues

Part of #10433 (no closing keyword: a stack closes it). Stack: #19833 (merged) → this PR → #21259 health → #21256 git → #21257 Sprites/boxd → #21258 Fargate → #21260 Kiro Cloud.

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

Contribution License Agreement

Contributions are accepted under the project's Apache-2.0 license; see the Licensing section of CONTRIBUTING.md.

@kyleseaman
kyleseaman requested review from a team and KiroCrewCaptain as code owners October 19, 2026 17:06
@kyleseaman
kyleseaman requested review from buluoray and removed request for a team October 19, 2026 17:06
@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Oct 19, 2026
@github-actions

github-actions Bot commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of 0fad680f2ce940dd00d872426db45d90973bfb9b — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

Design-Verdict: PASS

Nothing blocks. All new behaviour sits behind the developer flag, which is off by default. The PR includes 36 screenshots. Removing test_the_guest_has_no_write_mode is justified: that test recorded a missing caller, and this PR adds the caller.

[DESIGN-REVIEWED] 0fad680

Description read

[DESCRIPTION-READ] c64f132877143aebecddb203d13784bfa955112b68540cb18dc528152ec29683

That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 0fad680f2ce940dd00d872426db45d90973bfb9b — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

First-Principles-Verdict: PASS

Nothing blocks. Every surface sits behind the developer flag, which is off by default, and the new default_profile defaults to "" (this machine). So no one gets a different default without choosing it.

Notes

  • Item 6 — symptom-level: the guard only fires when the client sends ?slot=. Twelve other routes in routes/taskrunner.py that read this machine's files stay unwrapped: file-download, the three file-office-*, file-search, path-complete, file-grep, file-sheet, browse-dirs, browse-files, project/git/repos and project/tree. The description says it is scoped to "file, diff and Git", so this is a deferred point patch, not a blocker.

What this change ships

Inventory (10 items) — 9 justified

Intent: let the owner bind a chat to a Coder workspace and work on that workspace's files and terminal from the dashboard. ADDITION.

  1. A composer chip and a "Run this chat on" picker bind a chat to a profile — justified
  2. A Session Environments Settings tab for the Coder URL, profiles, default profile and stop switch, behind a Feature Previews switch — justified
  3. A new owner chat binds to the default profile when one is set — justified
  4. A chat bound to a profile is never prewarmed — justified
  5. A remote chat's side-panel terminal is a coder ssh shell in its workspace, and the gateway refuses a local one — justified
  6. File, diff and Git routes refuse a remote chat that names its ?slot= (409 chat_runs_remotely) — symptom-level; 12 sibling local-file routes are unguarded
  7. A Remote Files panel to browse, read and save files in the workspace, with conflict handling — justified
  8. The config editor and change cards validate session_environments with the loader's normalizers, a new normalized schema type with 4 consumers — justified
  9. Audit lines for these structured writes record the path only — justified
  10. An outdated guest worker is reinstalled once and the request retried — justified

[FIRST-PRINCIPLES-REVIEWED] 0fad680

Description read

[DESCRIPTION-READ] 6f58c66fbd3d7eddcb2ba0f52c3aaee452e4062b4e3f4fe663f749eff3313a71

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@github-actions

github-actions Bot commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5) — ✅ PASS

UX-level review of 0fad680f2ce940dd00d872426db45d90973bfb9b — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

UX-Verdict: PASS

Nothing blocks. The screenshots cover every control the PR adds, and the blind reader read each main-path control correctly. Those are the "Run this chat on" menu, Add Coder profile, Check Coder connection, the Files panel and the conflict actions. The reader declined Terminal only because they wouldn't know what to type.

Notes

  • sessionEnvironments.intro calls the Coder workspace "the machine where its files and commands live", but "This machine" in the picker means the local computer. The blind reader flagged that "machine" means two different things. Change the intro to "…its own Coder workspace, where its files and commands live".
  • The profile row's grey suffix shows the template (gpu-dev), while the delete warning names the ID (Coder GPU (ID gpu)). The reader couldn't tell "which one matters where". Label the suffix in the row, for example "template gpu-dev".
  • In the Workspaces list, the workspace name links to the chat, but the reader "can't tell whether they open the chat, the machine, or the Coder website." Put the link on the "In use by {{title}}" text instead, and leave the workspace name as plain text.

[UX-REVIEWED] 0fad680

Description read

[DESCRIPTION-READ] 7e4b9a8848786d7dd6b393f2d67ae470b7ef2d13b76008715cc01e10bca7cbd5

That sha256 names everything this verdict read: the PR description, plus the 39 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed 0fad680f2ce940dd00d872426db45d90973bfb9b — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 0fad680

Verdict parsed from the review's SHA-scoped output markers for commit 0fad680f2ce940dd00d872426db45d90973bfb9b.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 0fad680f2ce940dd00d872426db45d90973bfb9b false-positive: <reason, 20 to 500 characters>

@github-actions

github-actions Bot commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings (all downgraded on adjudication)

GPT 6.1 flagged blocking issues on 0fad680f2ce940dd00d872426db45d90973bfb9b; Opus 5.5 adjudication downgraded every one of them. Downgraded: the remedy each one needs is disproportionate to the harm it prevents. Adjudication downgraded all 1 blocking finding(s): the remedy each one requires is disproportionate to the harm it prevents.

This comment is updated in place on each push.

Review details

BLOCKING -- website/src/hooks/usePanelDocumentActions.ts:73 -- Chat-specific refusals poison the shared file-read cache

queryFn: ({ signal }) => fetchFileRead(filePath, signal, slot),
Trigger: Open a readable gateway file link in a remote chat, then open the same path in a local chat within ten seconds. The remote request receives 409, but fetchFileRead resolves normally and openFile caches that result under the slot-independent fileReadQueryKey(filePath).
Outcome: The local chat reuses the cached refusal without making its permitted request, reports HTTP 409, and does not open the file.
Anchor: defect
Fix: Include the requesting slot in the read-cache identity, distinguish chatless reads, and update the shared readers and invalidation.
Sites: website/src/hooks/usePanelDocumentActions.ts:116; cache readers at website/src/components/library/SessionDocPreview.tsx:62, website/src/pages/chat/SidePanelTabBodies.tsx:259, and website/src/pages/chat/page/eventBridges.ts:284; key and eviction at website/src/utils/fileReadQuery.ts:53 and website/src/components/useMarkdownPanelDisk.ts:196.

[BLOCK-MERGE-DOWNGRADED] 0fad680
[GPT-REVIEWED] 0fad680

Adjudication (Opus 5.5) — is blocking on each finding proportionate?

F1 (LOW): Every condition needed is confirmed in code. The developer flag has to be on (remote_chat_guard.py:66-68). The caller has to be the dashboard owner (:71-72). A remote chat has to request a gateway file even though the dashboard already hides those tabs for it (remote_chat_guard.py:5-7). And the same path has to be opened in a local chat within 10 s, because the entry goes stale after FILE_READ_STALE_MS = 10_000 (fileReadQuery.ts:49).
When it happens, the user sees an explicit HTTP 409 error and no tab opens (usePanelDocumentActions.ts:80-83). Nothing is lost. Retrying after 10 s fetches again with the local slot (staleTime, :74) and the file opens.
The real fix costs more than that: a slot-scoped cache identity has to be threaded through six readers plus key and eviction (fileReadQuery.ts:53, useMarkdownPanelDisk.ts:196, SessionDocPreview.tsx:62, SidePanelTabBodies.tsx:259, eventBridges.ts:284), and every future reader of the shared key carries it forever. That outweighs a self-correcting, visible, 10-second glitch on a developer-flagged path.

[ADJUDICATION] 0fad680f2ce940dd00d872426db45d90973bfb9b total=1 uphold=0 downgrade=1
DOWNGRADE F1 website/src/hooks/usePanelDocumentActions.ts:73 reason=disproportionate-remedy
[GPT-ADJUDICATED] 0fad680f2ce940dd00d872426db45d90973bfb9b

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 0fad680f2ce940dd00d872426db45d90973bfb9b false-positive: <reason, 20 to 500 characters>

@kyleseaman
kyleseaman force-pushed the feat/session-environments-coder-ui branch from 1a7f2c3 to e5722a8 Compare October 19, 2026 17:26
@github-actions github-actions Bot added merge conflict Branch has merge conflicts with its base — author must resolve before merge readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Oct 19, 2026
@kyleseaman
kyleseaman force-pushed the feat/session-environments-coder-ui branch from e5722a8 to cb93ffe Compare October 19, 2026 18:25
@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: action required A blocking check or review needs attention labels Oct 19, 2026
@kyleseaman
kyleseaman force-pushed the feat/session-environments-coder-ui branch from cb93ffe to 4563717 Compare October 19, 2026 18:28
@github-actions github-actions Bot removed the merge conflict Branch has merge conflicts with its base — author must resolve before merge label Oct 19, 2026
@github-actions github-actions Bot added the merge conflict Branch has merge conflicts with its base — author must resolve before merge label Oct 19, 2026
@kyleseaman
kyleseaman force-pushed the feat/session-environments-coder-ui branch from 4563717 to 062ebbb Compare October 19, 2026 19:01
@kyleseaman

kyleseaman commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=35610d3d35f4

Outcome: Kept, not changed.

remote_slot and slot are not two spellings of one request. remote_slot asks for the workspace shell and is refused when the chat is not remote; slot only scopes a LOCAL shell so the gateway can refuse it for a remote-bound chat. Folding them into one parameter that silently turns a local tab into a remote shell would change what an already-open tab runs on. The stacked PRs (#21259 onward) also open terminals through remote_slot, and Kyle asked that B's public surface stay stable while they sit on it.

@kyleseaman

kyleseaman commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=ae2694e399d6

Outcome: Kept, not changed.

The leaf paths have a production writer: change cards. Every _EDITABLE_CONFIG path is card-editable, and change_card_catalog.py (added in this PR) classes them: default_profile and stop_on_shutdown are scalar settings a setting.change card can write (cannot-widen), and coder_url, profiles and the whole section are classed code-exec, since they pick the machine that runs a session's commands. The settings registry deep-links the same keys.

@kyleseaman

kyleseaman commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=129c8e0762b0

Outcome: Kept, not changed.

The client that omits keys is this PR's own card once the stack lands: #21258 (Fargate), #21257 (Sprites/boxd) and #21260 (Kiro Cloud) each add their own profile list to SECTION_OPTIONAL_KEYS. A browser still holding this card's bundle after the gateway updates saves the section without those keys, and without the merge that save deletes every Fargate/Sprites/Kiro Cloud profile. The merge runs under the config write lock, so it costs one dict copy.

@kyleseaman

kyleseaman commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=a2cd92517031

Outcome: Fixed in the description (head 062ebbb).

The description now says the Settings tab sits behind a per-device Feature Previews switch (mc-preview-session-environments) that this PR adds, instead of "as before".

@kyleseaman

kyleseaman commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=04b28be71f4b

Outcome: Fixed in 062ebbb.

The switch now names what it controls: "Let Kiro Crew stop this profile's workspaces". The hint says when to leave it on (the template keeps the working folder across a stop, so the files are still there) and when to turn it off (a template that loses its files, such as a MicroVM; Kiro Crew then never stops it). All 12 catalogs.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=d2ba500dcd38 terminal.py:557: In-flight local terminals survive a switch to a remote profile

Outcome: Fixed in 960a423.

The opener now reads the placement again once its shell is published: after the open/reconnect block in api_terminal_ws, a session with a chat_slot re-runs _local_terminal_refusal and, on any refusal, calls retire_local_shells_of_chat with that refusal's message and code before the read loop starts. Publication happens before that read and the bind writes its choice before its retire scan, so either the re-check sees the new profile or the retire sees the published shell. The test test_a_shell_still_opening_when_its_chat_is_bound_is_closed binds the chat right after the first placement check of a real local shell and asserts the socket gets terminal_wrong_machine and the shell is gone from the registry; it fails without the re-check.

@kyleseaman

Copy link
Copy Markdown
Contributor Author

Inherited red on 960a42315c: macOS Tests / Backend Tests (macOS) (5) fails only test/test_ai_review_workflows_human_override.py (every case), with ${ACTOR,,}: bad substitution from the step script. That lowercase expansion is bash 4 syntax and comes from .github/workflows/ai-review-human-override.yml, last changed on main by #22698 (1a0924b53a); macOS runs bash 3.2. This PR changes no file under .github/ and nothing that test reads. The same failure hit the previous head f16fffd53f (job 120757401347) and this one (job 120771657280). Re-run once; every other check on this head passes.

@iamwhatever

Copy link
Copy Markdown
Contributor

HOLD at 960a423.

  • This is reserved for a human reviewer. It adds seven dashboard routes, including a remote workspace file write, and it changes which machine a chat's terminal runs on. Every new handler is owner-gated.
  • These lane notes on this head have no reply:
    • Opus: terminal.py is now 1,749 lines (main has 1,553), past the 1,600-line split point. r119 and r123 raised this too.
    • Design: with the flag off, every owner chat create still loads the config for apply_default_environment. Return early when the flag is off, or say why not.
    • First Principles (Item 9): PREVIEW_SESSION_ENVIRONMENTS is a second gate on top of the developer flag.
    • UX: the workspace rows show no chat title, the Settings rail does not show the active tab, and the "Default for new chats" help text is unclear.
  • "Before the flag is turned on" lists a live Coder run and a gateway-side guard for the file, diff and Git routes. Neither is tracked. Open an issue for each and link them.
  • The on-demand macOS shard 5 is red on this head. The author says main's workflow causes it; a same-test red on main would confirm that.

The conflict from r123 is gone. Opus and GPT have no blocking findings, and the earlier findings have dispositions.

— crew-pr-reviewer (r124)

@kyleseaman

Copy link
Copy Markdown
Contributor Author

Rebased onto main 085ef5678e (head 2ac56e2150). One non-mechanical conflict, in website/src/pages/chat/SidePanel.tsx: main #21325 keeps the local Files tab mounted (hidden) when another tab is selected, and removed the tab's if (!isActive) return null. This PR's remote branch sat after that early return, so the return now lives inside the if (remoteEnvironment) block. Local Files follows #21325 unchanged; the remote RemoteFilesPanel still unmounts when inactive, as before (it has no active prop to pause its guest reads, so keeping it mounted would keep polling the workspace in the background). No other change to this PR's patch.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=48c4260c37cd terminal.py:531: a dock terminal can never reconnect to its live shell after its chat is closed or deleted

Outcome: Fixed in fa32704.

_local_terminal_refusal takes the live session a reconnect would take over (reattach=). When the chat is gone, the dial goes through only for that same session id's shell, and only while it is live, local, and recorded for that chat. api_terminal_ws marks such a dial reattach-only: if the shell exits or is replaced before the takeover, the dial is refused with terminal_placement_unavailable and a denied audit record, before any new shell spawns. The post-publish re-check passes the same session on a reconnect. Binding a chat still retires its local shells. Nothing changes with the dev flag off. Tests: test_a_terminal_reconnects_to_its_shell_after_its_chat_is_closed (end to end: close the chat, redial, the same shell answers), test_a_gone_chat_lets_only_its_own_live_local_shell_reconnect (the rule, with a remote, other-chat and dead shell each refused) and test_a_gone_chats_shell_dying_mid_dial_opens_no_new_local_shell (counts PTY spawns). All three fail without the fix.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=f2de2e884b06 test_session_environment_handlers.py:685: Terminal retirement test rejects normal PTY output

Outcome: Fixed in fa32704.

Both retirement tests now read the error through one helper, _error_code_of. It skips BINARY frames (shell output) and other control frames under a 30-second deadline. A close before the error frame still fails, as does a missing error, and the code is still asserted. test_the_error_frame_reader_skips_shell_output feeds a prompt frame and a ready frame before the error, and checks that a premature close still fails. It fails without the skip.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=335015659092 RemoteFilesPanel.tsx:265: A successful refresh strands unsaved edits

Outcome: Fixed in c399f47.

When the selected file has a retained draft and its read is not editable (binary or truncated), the file header now shows "Copy my edits" beside "Discard my edits", together with its own copy-failure notice. Save stays unavailable, and while a save conflict is shown, its own copy button is used instead. The leave-page guard stays armed until the draft is discarded. keeps unsaved edits copyable when a refresh finds the file grows past the read limit and ... becomes binary edit a file, refresh into each read, copy (one failure, then success) and check Save is absent. Both fail without the button. Screenshots are in the PR body under Remote Files.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=f2de2e884b06 test_session_environment_handlers.py:833: Real PTY tests leave shells running on failure

Outcome: Fixed in c399f47.

All four real-PTY tests now end in _close_with_shells(client, app), which pops every session still published in the app's terminal registry and stops it with _kill_session (child process, then PTY descriptor and reader) before it closes the client. A failed assertion therefore leaves no shell or blocking reader behind. test_a_failing_pty_test_leaves_no_shell_running publishes a real shell, raises inside the try, and checks the registry is empty and the process is no longer alive. It fails when the cleanup is reduced to client.close().

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=86880e645bcc session_environments.py:421: Default binding leaves already-open gateway terminals alive

Outcome: Fixed in 7114829.

apply_default_environment now takes the terminal registry. Once it records the default choice, still under the session's switch lock, it calls retire_local_shells_of_chat for that chat, as the explicit switch route does. A local shell another tab opened in the gap between the chat's publication and the bind therefore gets the wrong-machine code and is stopped. api_chat_slot_create passes state._terminal_sessions. Workspace shells and other chats' shells are left alone, and a create that binds nothing retires nothing. test_a_default_bind_retires_the_chats_local_shells fails without the retire.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=86880e645bcc session_environments.py:160: Registry read failures masquerade as an empty workspace inventory

Outcome: Fixed in 3de0b7d.

api_session_environments_workspaces_get no longer turns a registry OSError or WorkspaceRegistryCorrupt into rows = (). It returns HTTP 500 with code: "workspaces_unreadable" and a fixed message, so no host path reaches the browser, and logs the detail. An empty list now only ever means a successful read with no bindings.

The endpoint has one consumer, SessionEnvironmentsCard (the Manage list and the removed-profile check share the query), and the card never renders the server text. The Manage list already shows its localized "The workspace list could not be read." notice. The removed-profile check used to read a failed query as zero stranded chats. It now says, through the new i18n key sessionEnvironments.removedInUseUnknown in every locale, that it could not check which chats still use the removed profile.

Tests: test_workspaces_endpoint_lists_owned_rows_and_reports_an_unreadable_registry covers a corrupt file and an I/O failure, each returning the coded 500 with no path. It fails with the old rows = (). says it could not check a removed profile's chats when the workspace list fails fails without the new note. Screenshots are in the PR body.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=86880e645bcc session_environments.py:440: Default binding leaves an in-flight local prewarm alive

Outcome: Fixed in ec5649d.

Under the same slot_switch_session_lock(key), and before it records the default choice, apply_default_environment now runs the explicit switch route's sequence with no new state:

  • it cancels the slot's _eager_spawn_task if one is still running, and drains it with asyncio.gather(..., return_exceptions=True);
  • if a runtime is registered and the slot had an eager task, it removes that runtime with sessions.remove(key);
  • it clears a _pending_reset_history_key armed for the key.

A runtime registered with no eager task came from a first turn, so the chat stays local, as before. A prewarm started by another tab's project change therefore can't register a local runtime that the first message would reuse.

Tests: test_a_default_bind_stops_a_local_prewarm_still_starting holds a prewarm mid-handshake across the bind, then lets it finish. It checks that the bind cancelled it, that no local runtime registered, and that session_runs_remotely is True for the chat. test_a_default_bind_removes_a_runtime_its_prewarm_registered covers a prewarm that already registered. Both fail on the previous bind, and test_a_runtime_a_first_turn_registered_keeps_the_chat_local keeps the first-turn case local.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=99f6356c924c ui-index.generated.json:371: Record the preview prerequisite before offering settings guides

Outcome: Fixed in b181f79.

PREVIEW_FLAG_ENABLERS now maps PREVIEW_SESSION_ENVIRONMENTS to developer.session-environments, and SETTINGS_TAB_PREVIEW gates the session-environments tab, matching SettingsPage's buildTabs filter. The generator (scripts/lib/ui-index.mjs) now gives every setting and Settings sub-page its tab's preview requirement, and the index is regenerated. The tab and its four settings carry preview_flag with the Developer switch. find_ui("coder url") then answers the request in words with the preview note, instead of offering a settings.show guide into a hidden tab.

Scope: on a clean main checkout, the generator change alone leaves ui-index.generated.json byte-identical (the webhooks tab has no settings), so no other tab's entries move. A second gen:ui run makes no change.

Tests: a generator unit test (a gated tab's setting and sub-page inherit its preview), a committed-index test (the tab and every session-environments.* setting require the preview; the Developer switch does not), and test_a_setting_on_a_preview_gated_settings_tab_is_answered_in_words. All three fail when the descriptors, generator and index are reverted.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=673050e96414 core_config_schema.py:481: setting session_environments.default_profile alone stores an id that matches no saved profile

Outcome: Fixed in 6ca0f8d.

The leaf write in api_kirocrew_config_patch now checks the value against the saved section's profile ids inside _mutate_config_patch, while holding the config write lock (_check_default_profile_is_saved). An unknown id is refused with HTTP 400 code: "unknown_profile" and nothing is written. Change cards apply through the same PATCH /api/config/kirocrew, so they are covered too. "" and null (this machine) are always accepted. Whole-section saves are unchanged: one save may add a profile and make it the default, and deleting the default profile drops the default, never a 400.

Tests: test_the_default_profile_leaf_names_only_a_saved_profile and test_a_default_profile_leaf_with_no_saved_section_is_refused both fail without the check. test_one_save_adds_a_profile_and_makes_it_the_default covers the add-and-default and delete-the-default cases. The refusal message carries no path.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=86880e645bcc session_environments.py:36: Placement and file reads use different config snapshots

Outcome: Fixed in 6ca0f8d.

_profiles() now reads the applied live.snapshot(), the same snapshot browse_session's _resolve and a remote start use, and falls back to KiroCrewConfig.load() only when no snapshot exists. Placement therefore never advertises a working folder the file browser has not applied yet. test_the_profile_list_follows_the_applied_snapshot_not_the_disk fails when _profiles() reads the disk again.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=a62dfd9054c0 SessionEnvironmentsCard.tsx:204: Editing settings hides an in-flight reopen failure

Outcome: Fixed in 6ca0f8d.

ReopenChat now shows "Could not reopen that chat." whenever the reopen fails, whether or not the card holds a draft. Only the "Ask the agent" hand-off is withheld while a draft exists (askAgent={!blocked}), since leaving the card would lose the unsaved edits. says a reopen failed even when edits started while it was running fails a pending reopen after a draft starts, and the updated says so when an archived chat cannot be reopened… keeps the notice once a draft exists. Both fail on the previous code.

@iamwhatever iamwhatever left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

REQUEST_CHANGES on 6ca0f8d. This also stays reserved for human review: it adds seven dashboard routes, including a remote workspace file write, and changes which machine a chat's terminal runs on. Every new handler is owner-gated, and the Opus and GPT findings since r124 all have fix dispositions.

What needs changing:

  • The two "Before the flag is turned on" items (a live Coder run, and a gateway-side guard for the file, diff and Git routes the client alone keeps off a remote chat) are still untracked, as r124 noted. Please open one issue for each and link both in the description.
  • The branch conflicts with main in test/test_change_card_setting_classes.py. Rebase.
  • Lane notes on this head still have no reply: First Principles item 9 (PREVIEW_SESSION_ENVIRONMENTS as a second gate), the UX notes (strand warnings name profiles by id, duplicate intro text, Copy/Discard over a binary file, the cut-off 256 KB label), and terminal.py at 1,805 lines, past the 1,600-line split point. A fix or a one-line reason for each is enough.
  • macOS backend shard 3 is red on this head. The description blames leaked leases that #24590 fixes; a link to the same failure on main would confirm it.

A new head after the rebase needs a fresh review.

— crew-pr-reviewer (r128)

@kyleseaman

kyleseaman commented Oct 27, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=492e113e5b07 SidePanel.tsx:960: Placement refresh failures discard unsent document feedback

Outcome: Fixed in f69ce3c.

Placement gating is unchanged: a file tab is still withheld, and no gateway action is available, until placement is read again. What changed is that the unsent "overall instruction" now survives that unmount. CommentList keeps it in an in-memory map, never written to browser storage or sent anywhere, under a key from instructionDraftKeyFor. MarkdownPanel passes the tab's slot-scoped identity (slot + tab id) with the gateway-local path, so the same file open in another chat never shows this chat's text. A read-only wtdiff panel, or one with no tab identity, keeps nothing. The draft is cleared on Submit All, when the instruction box is closed and when every pending comment is removed.

Tests in CommentListInstructionDraft.test.tsx cover: survives an unmount in the same chat; never shown in another chat; cleared on submit; cleared on close; cleared when the comments are gone; never written to storage. Each condition fails when its part of the fix is reverted.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=492e113e5b07 SidePanel.tsx:960: Failed placement refresh discards unfinished inline comment edits

Outcome: Fixed in 72a92db.

Placement gating is unchanged: the tab is still withheld until placement is read again. An inline edit of a pending comment that is open when the panel unmounts now survives it. CommentRow keeps the typed text in an in-memory map (rowEditDrafts in CommentOverlay.tsx), never written to browser storage or sent anywhere. The key is the same chat-scoped instructionDraftKeyFor key plus the comment id, so another chat with the same file open never sees it, and a list with no key keeps nothing. On remount the row reopens in edit mode with the typed text, and Enter or blur saves it as before. The kept edit is dropped on save, Escape and Remove, and when its comment is no longer pending. The :976 site (the refresh turns remote) goes through the same unmount and is covered by the same fix.

Tests in CommentListInstructionDraft.test.tsx cover: the edit reopens after an unmount and saves on Enter; it is not kept after save, cancel, in another chat or with no key; it is dropped when its comment is gone; it is never written to storage. Each fails when its part of the fix is taken out.

@kyleseaman

Copy link
Copy Markdown
Contributor Author

Replies to the lane notes and the review asks, as of 72a92db210:

First Principles, second gate (mc-preview-session-environments). Kept on purpose. Part A already gates the runtime on the server. This per-device Feature Preview only hides the unfinished Settings surface (the tab, its search rows and guide entries), the same pattern as PREVIEW_WEBHOOKS. Without it, every dashboard user would see the tab before the preview graduates.

UX, terms. No rename in this PR. Each term names a different thing: Session Environments is the page, a profile is a saved setup, a workspace is the Coder machine each chat gets, a template is the Coder template a profile names, and "this machine" is the gateway host. The intro defines profile and workspace where the reader starts.

UX, fixed in 72a92db210:

  • Removal warnings name a profile the way the list shows it, with its ID: "…can't run until Development (ID dev) is added back." A profile with no name is still named by its ID.
  • The repeated "A profile is a saved setup…" under Profiles is gone. That hint now says only what the intro doesn't: a profile is a template plus a working folder, and where to pick a chat's profile.
  • The 256 KB note no longer rides on the truncated path. A file that is too large or binary says why on its own line. With unsaved edits it says what to do: "This file is now too large to edit here (only the first 256 KB is shown). Copy your edits to keep them." The binary state has its own wording.
  • All 12 catalogs are updated.

terminal.py at 1,805 lines. Split in this PR with a behaviour-neutral move. The placement block goes to a new owner, terminal_placement.py (170 lines). The completion listing helpers (_list_completions, _list_vetted_completions, _resolve_vet_and_list) go to their existing owner, terminal_complete.py. The facade is now 1,534 lines. Every terminal.<name> read and patch still works through the facade mirror. The only edits besides the move: black wrapped one long return in the new module, and one docstring example dropped the brand spelling.

macOS shard 3. The cap is inherited from main. Main's nightly run 40459526858 on a796ca5224 was cancelled at the same 40-minute cap on macOS 1, 2 and 3. The gc-drain change in #24629 is the likely cause. #24590 was closed unmerged. On f69ce3ca31, Linux 2 and Windows 2 failed only on test_a_warm_session_read_that_earns_a_savepoint_reads_only_the_gap: it is marked as expected to fail, now passes, and strict mode turns that pass into a failure. That is also inherited, and open #24774 fixes it.

The rebase is mechanical, and test/test_change_card_setting_classes.py has no conflict left.

@kyleseaman

Copy link
Copy Markdown
Contributor Author
  • span=4464213526e3 BLOCKING — docs/feature-map/README.md:152

Outcome: Fixed in db0a670.

The Terminal panel row now says a chat bound to a session environment gets its per-chat terminal in its Coder workspace instead (?remote_slot=, a coder ssh shell), that a local shell for that chat is refused (terminal_wrong_machine), and that its local shells are closed once it is bound. The backend column lists handlers/terminal_placement.py (which machine a chat's shell runs on) beside handlers/terminal.py. The feature-map gate passes against this base.

@kyleseaman

Copy link
Copy Markdown
Contributor Author

UX Review on badfce5acd, as of db0a670cbc:

Blocker, "Working folder" gives no example: fixed. workdirHint now reads "Full path of the folder in the workspace where the chat's files and commands run. It starts with /, for example /workspace/project." (the backend accepts only a path starting with /), in all 12 catalogs. SessionEnvironmentsCard.test.tsx checks the new hint and fails on the old text. The "Profile editor" screenshots in the description show it.

Note, removedInUse / removedInUseUnknown name the bare ID: stale. Since 72a92db210 both name the profile as the list shows it, with its ID: "…can't run until Coder GPU (ID gpu) is added back", "Could not check which chats still use Coder GPU (ID gpu)". The screenshots this review read predated that; every card and Remote Files screenshot whose copy changed is now re-shot from this head ("A profile just removed", "Removed profile, workspace list unreadable", the card, check-failed, empty, editor, picker, setup, saved, save error, and the truncated / binary Remote Files shots).

The other notes (defaultMoved wording, a Retry on the workspace-list error, labelling the workspace link) are not blockers and are not changed in this PR.

@iamwhatever iamwhatever left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

REQUEST_CHANGES at db0a670. This also stays reserved for human review: it adds seven dashboard routes, including a remote workspace file write, and changes which machine a chat's terminal runs on. Each new handler is owner-gated (session_environments.py).

The r128 lane notes are answered: the preview switch has a reason, the profile names and 256 KB copy are fixed, terminal.py is split to 1,534 lines, and the macOS shard is linked to main's own run.

What needs changing:

  • The two "Before the flag is turned on" items (a live Coder run, and a gateway-side guard for the file, diff and Git routes) are still untracked. The Design Review on this head repeats the second one. Please open one issue for each and link both in the description.
  • The branch conflicts with main again. Rebase.
  • The UX notes on this head have no reply: raw backend strings shown as-is, workspace rows that don't name their chat or profile, and the "Run this chat on" explanation repeated three times. A fix or a one-line reason for each is enough.

A new head after the rebase needs a fresh review.

— crew-pr-reviewer (r129)

The dashboard half of remote session environments, on top of the Coder
backend: the Settings card with its readiness check and workspace list,
the composer environment chip, the remote Files panel and terminal, the
side panel's placement guard, the per-chat environment and file routes,
default-profile routing for new chats, and the config-editor validators.

Behind the developer flag (off by default): while KIROCREW_DEV_MODE is
unset the gateway offers no profile, so the picker stays hidden and a new
chat is never bound to the default profile.
@kyleseaman

Copy link
Copy Markdown
Contributor Author

Replies to r128 / r129, as of 0fad680f2c (rebased onto 30023be4d1; two commits so the UX delta reads on its own):

The two "Before the flag is turned on" items are handled in this PR, so neither needs a tracking issue.

  • Gateway-side guard for the file, diff and Git routes: built in 2108437011. /api/file-read, -raw, -stream, -watch, -write, -diff, /api/project/git, /git/status and /git/log refuse a request whose ?slot= names a chat on a session environment (409 chat_runs_remotely, one fixed message that names no path; an unreadable placement counts as remote). A request without ?slot= passes unchanged, the routes stay owner-only, and a non-owner gets the route's own answer. With the flag off nothing is refused. The side panel's open, diff and save send their chat's slot=. Code: src/kiro_crew/dashboard/remote_chat_guard.py, wired in dashboard/routes/taskrunner.py; tests: test/test_remote_chat_file_guard.py, usePanelDocumentActions.chatSlot.test.ts.
  • Live run: Kyle (PR owner) has run the Coder environment end to end against a live Coder deployment.

The description's Manual verification section now says both.

Rebase: mechanical. The ui-index is regenerated, and the conflict with #24091 in ChatPane.tsx and activeSlotState.ts was adjacent lines, resolved by keeping both sides.

UX notes on db0a670cbc, fixed in 0fad680f2c:

  • Raw backend strings. A refused Coder URL now says "Enter the full address of your Coder deployment, for example https://coder.example.com." under the box, and a failed template list says "Couldn't load your Coder templates, so type the template's name." In both, the gateway's or Coder's own words sit folded under Details, the same fold the readiness check already used, because they are needed to diagnose.
  • Workspace rows don't name their chat or profile. Each row now shows "Profile: ", and a live row says "In use by ". Both come from what the page already holds (saved profiles, the sidebar's chat list), with no new API field. A row whose title is not on hand still says "In use by a chat".
  • "Run this chat on" explained three times. Where the menu is is now said once, under Profiles; the Manage paragraph keeps only the archiving sentence.

Each fix has a test that fails without it (4 in SessionEnvironmentsCard.test.tsx). All 12 catalogs are updated, and the affected screenshots are re-shot from this head.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants