Live m4air run: a signed-in owner whose active team never had a trial was sent
to T3/Checkout. Only teams bound in the browser got the trial. Session creation
now runs the same grant (N11 limits) for the active team, so the chain skips
the plan page.
- One row from the Orca account, never the sandbox: 32 px ready line when every agent is ready at
mount, otherwise 48 px; every later state renders inside that height (Playwright: measured layout
shift 0, logs toggle never moves).
- R2 Set up runs S0's setup inline for each agent in turn; device code, browser, paste, failed,
expired, kill switch (per-workspace sign-in, queued until the sandbox answers), A6 desktop key.
- ⋯ menu: Source per agent in this workspace (new per-workspace setWorkspaceSource / workspaceSources
IPC; main resolves the sandbox id from its own runtime records), Manage agent accounts.
- Removes the v3 'Add an agent' build step and the per-sandbox Cloud sign-in row from the sheet.
The desktop parsed the DELETE answer with the full strict sandbox view, so any view field the control
plane adds later (e.g. archive state) turned a successful delete into 'Orca Cloud returned an unexpected
response'. Destroy now parses only what confirms the delete (id + status) and ignores unknown keys; a
body without them is still rejected. Tests cover both.
The merged Set up Orca Cloud panel's 'agent' step now comes from the same
row model as the standalone row: done with 'Claude · from your Orca account' /
'Codex · team key', Set up once / Renew to S0, the kill-switch info line,
signed-out copy, loading; hidden when nothing there helps. Drops the panel's
placeholder copy keys.
The cause usually prints last, so the Details window now keeps the tail,
opens scrolled to the end, and notes '…N earlier lines' (plurals) at the
top. Copy details still copies the full text.
- setup-token on Windows goes through cmd with the shared quoting as one verbatim
ConPTY line; the child env is buildClaudeChildProcessEnv (no inherited auth) and
runs in the home directory.
- Token capture drops styling escapes, breaks on cursor moves, waits 1 s for a
repaint and keeps the longest complete copy, so a split copy is never stored.
- Broker ids must start with a letter or digit (no '.'/'..' path tricks), checked
at parse time too; server error names are looked up with a Map.
- A setup belongs to the Orca identity it started under: a changed identity drops
the captured token / cancels the broker and resets the rows.
- A token being stored can't be cancelled or raced by a paste or a new start.
- The Codex driver only shows OpenAI sign-in pages and caps the wait at 20 min.
- The paste field clears in a finally; a rejected IPC reads as unknown.
Local custom-recipe failure cards now show a plain-language summary with
the raw error in a collapsed Details disclosure (monospace, Copy details,
truncated to 20 lines with a note; Copy takes the full text). Orca Cloud
cards are unchanged. Adds collapsed/expanded/truncated fixtures and tests.
Main:
- agent-credentials client for /v1/agent-credentials, /v1/agent-sources and the
Codex broker routes; replies and errors are codes only. The shared control-plane
request keeps the server's error name (HostedSandboxRefusalError) so newer
routes (broker_in_progress, rate_limited, no_golden_image) map to codes.
- claude setup-token runner (pseudo-terminal in production, fake child in tests):
captures the sk-ant-oat01 token in memory, never logs it, PUTs it from main
with the local Claude account email as label. Capture failure or no claude
offers "Paste the token instead"; a pasted value goes renderer -> main once.
- Codex broker driver: device code, poll until active, cancel, code expiry.
- orcaCloudAgents IPC + preload bridge: states and codes only.
Renderer:
- Settings > Agent accounts > Orca Cloud (S0, spec §5): not set up, setting up
(browser / preparing / device code / paste / saving), state 7 Source choice,
active, expiring, expired/revoked, kill switch, signed out, load and save
errors; the mandatory policy line; Source via SettingsSegmentedControl.
- The composer's Agent row (cloud-start G) replaces ComposerCloudAgentHint:
from your Orca account / team key (collapsed), Set up once -> S0, kill switch
info, signed-out copy; hidden for members blocked by billing; never blocks.
- Telemetry cloud_first_start_agent_coverage (once per install).
- Fixtures: settings/agent-accounts/OrcaCloudAccounts:*, new-workspace/OrcaCloudAgentSetupRow:*.
design-squad, after the W6 live run: an unknown asker reads 'A command asked for NAME' with no repeated 'from' line; the saved detail no longer claims the agent never gets the value (it is a plain env var); the CLI notes say the same. The saved card is announced, then dismisses itself after 6 s (paused while hovered or focused); while asking, scrolling or typing in the pane folds it to the pill.
- Signed in but Orca unreachable: rows keep their last-known state (from the per-account ledger:
names, key ids and timestamps only) under "Can't reach your Orca account right now", with no
account actions that would fail.
- Saves made while signed in that miss the account (offline or a transient failure) are queued in
the ledger and retried at start, on wake and with backoff while waiting; the timer stops when
idle. Only that key at that version: edits, removals and signed-out saves are never uploaded.
- Fixtures: removePrompt, cloudUnreachable.
Set up and start opens the one-tab setup session for the project's GitHub repo.
The setup panel reads each poll's snapshot. A draft restored after a restart
resumes polling its session.
- use-wake-attach-hold: only visible time counts toward the 8 s cap (+ test)
- cloud-workspace-format: formatRelative via Intl.RelativeTimeFormat in the UI locale (+ test)
- share-port-dialog-parts: reserved ports from SHARE_RESERVED_PORTS (was missing 6080), MAX_PORT_CHOICES
- archived 'more files not shown' gets a _one form; localization catalogs synced
The boot resolves the session env before the VM is up, and a secret change only refreshes running sandboxes, so a value saved from the build-sheet Secrets row during the boot never reached the workspace (found in the W6 live e2e). Once running, the boot re-resolves the env and refreshes the session when it changed.
--wait is a global boolean flag, so '--wait 300' left 300 as a positional and the command failed with 'Unknown command: secret request NAME 300' (found in the W6 live e2e). A spec can now list optionalNumberFlags; secret request does for wait.