Preserve returned change request bodies as search-only option text so client filtering keeps description matches. Keep title and ID matches ranked ahead of secondary text.
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
Command-scoped user.name overrides the fixture repository configuration and breaks its author assertion. Supply the fixture author when creating commits and exercise the override with real Git.
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
Format projected text entries in sequence instead of grouping replies ahead of their preceding reasoning. Timeline projection already owns fragment merging.
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
The failing signer uses gpg.program, which Git bypasses when a global SSH signing format is inherited. Select OpenPGP in the commit and squash-merge test repositories so both checks exercise their configured failing signer.
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
Request uncolored patches with explicit a/ and b/ prefixes before parsing. Mnemonic paths do not match checkout entries, and forced ANSI colors hide patch headers and hunks.
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(server): claim scheduled slots atomically
* fix(server): skip elapsed schedule slots
* test(server): isolate overlapping tick runner
* test(schedule): reproduce duplicate cron dispatch across overlapping ticks
* fix(schedule): preserve final slots and recovery after failed outcome writes
Historical running records are not proof of a live execution after an outcome write fails. Keep live-run exclusion in the existing in-memory guard while the persisted cursor prevents duplicate slot claims. Final allowed runs do not need a future cron occurrence; recover interrupted final runs and preserve cadence if their limit changes during execution.
---------
Co-authored-by: Keith Kowal (paseo-dev) <keithpk+paseo-dev@kowal.me>
Co-authored-by: keithpk <keith@kowal.me>
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(cli): keep streamed message fragments together in logs --follow
`paseo logs` without `-f` fetches the timeline projected, so
`projectTimelineRows` merges consecutive assistant and reasoning fragments by
exact concatenation before the transcript is rendered. Follow mode never got
there: it called `formatAgentActivityTranscript([item])` per arriving event, so
every fragment was projected alone and printed on its own line. Words broke at
fragment boundaries, a reader could not tell an author newline from a transport
one, and `[Thought]` was reprinted for each reasoning chunk.
Follow mode now keeps a fragment chain open using the same merge rule the
projection uses (same kind, same `turnId`, matching `messageId` when the event
carries one) and writes the text as it arrives instead of one line per event.
Only trailing whitespace is held back, so a message still ends trimmed the way
the one-shot transcript ends it, and the chain closes when another kind of item
arrives or the follow stops.
* test(cli): place daemon follow regression in its owning suite
* fix(cli): preserve filtered log boundaries and run the follow regression in CI
* test(cli): wait for a completed log line before terminating follow
---------
Co-authored-by: Lorenzo Fiore <fiorelorenzo.fl@gmail.com>
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(gitlab): find merge requests and issues by number
GitLab text search matches titles and descriptions but not item numbers. Use repository-scoped IID filtering for numeric queries, keeping open-only results and the existing text search behavior.
* fix(gitlab): retain text matches in number searches
Supplement title and description search with the open-item number lookup, rather than replacing it. Prioritize exact numbers, deduplicate results, and preserve each list limit.
* refactor(gitlab): name list helper arguments
---------
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(app): decode percent-encoded assistant file links
markdown-it percent-encodes non-ASCII and space characters in link
hrefs, so a link like [x](docs/reports/报告.md) arrives as
docs/reports/%E6%8A%A5%E5%91%8A.md. The file:// and bare absolute-path
branches already run safeDecodeURIComponent, but the workspace-relative,
path:line, Windows, and home-relative branches did not, so clicking the
link tried to open the still-encoded name and failed with ENOENT.
Decode the path portion before the relative-path heuristics and in the
inline absolute/Windows branches so all href forms behave the same.
Generated with [Devin](https://devin.ai)
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(app): decode hrefs only, never raw inline-code paths
Raw inline-code and linkified tokens reach the same parser as markdown
hrefs. Decoding unconditionally would send a real file named
my%20file.md to my file.md. Gate every decode on AssistantHrefParseOptions.decodeHref — set by the resolver only when the token
came from source.href — and extract parseWindowsFileLink to keep the
entry point under the complexity limit. Adds parse-level and
resolver-level regression tests for literal %20 filenames.
Generated with [Devin](https://devin.ai)
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(app): keep raw absolute paths out of URL parsing
new URL() percent-encodes non-ASCII, so a raw inline-code path like
/tmp/开户.md became /tmp/%E5%BC%80%E6%88%B7.md when decodeHref was
false. Hrefs still go through URL parsing (normalizes dot segments
before decode); raw paths split the # fragment manually instead.
Generated with [Devin](https://devin.ai)
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(app): keep encoded and literal file links distinct
Preserve inline-code provenance for Windows paths and key cached file lookups by their resolved query, path, and line range. Reproduce the reported non-ASCII Markdown link through the renderer and filesystem, with browser coverage for clicking it.
Co-authored-by: Ouyang Xingyuan <galaxyroaming@qq.com>
Co-authored-by: dengjiawen <dengjiawen@hortorgames.com>
* test(app): assert file content after opening the markdown link
Co-authored-by: sunshinehomeyu-blip <267551119+sunshinehomeyu-blip@users.noreply.github.com>
* fix(app): preserve encoded line markers in filenames
---------
Co-authored-by: Ouyang Xingyuan <galaxyroaming@qq.com>
Co-authored-by: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
Co-authored-by: dengjiawen <dengjiawen@hortorgames.com>
Co-authored-by: sunshinehomeyu-blip <267551119+sunshinehomeyu-blip@users.noreply.github.com>
* feat(app): modernize compact headers and sidebar affordances
Use a shared island header with quick tab creation and keep sheet footers inside their visible layout. Opt sidebar rows into touch sizing without changing global appearance settings.
* test(app): match darker light-theme workspace selection
* test(app): align plugin journeys with compact header
Failed bridge requests discarded the daemon error body before OpenCode could show the validation explanation to the agent. Preserve that message and status while retaining the HTTP fallback.
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(cli): honor --since in agent logs
* test(cli): verify logs cutoff in the daemon-owned suite
Keep the real CLI and WebSocket regression while placing daemon fixtures
in the server suite, as the CI ownership contract requires. Invoke the
source CLI so the test runs with the server suite's dependency build.
Co-authored-by: 见安 <73627805+Mar-garet@users.noreply.github.com>
* test(server): use a portable loader URL for the logs regression
---------
Co-authored-by: 见安 <73627805+Mar-garet@users.noreply.github.com>
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix: prevent file mention autocomplete from triggering on emails
Skip @ when preceded by a word character or +, and reject angle
brackets in mention queries so email addresses no longer open the
file autocomplete or steal Enter.
Closes#1364
* test(app): verify emails send intact without file completion
The word-character and plus guard tests previously passed because their queries included trailing spaces. Exercise each guard with the cursor at the end of the email, and cover Enter in the browser alongside normal file completion.
Co-authored-by: Jason <mildreins@gmail.com>
---------
Co-authored-by: Jason <mildreins@gmail.com>
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(cli): pass cwd to worktree ls and archive
worktree ls/archive called getPaseoWorktreeList({}) which fails because
the daemon requires cwd or repoRoot. Default to process.cwd(), accept
optional --cwd, and cover both with unit and CLI help tests.
Closes#2010
* test(cli): reproduce worktree listing without repository scope
* fix(cli): resolve local worktree paths from the calling directory
Local instance paths belong to the CLI working directory. Explicit endpoint paths remain daemon-relative, matching the project command convention. Cover both through the real daemon.
* refactor(cli): make repository directory selection explicit
---------
Co-authored-by: Jason <mildreins@gmail.com>
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix: resolve real base branch for worktrees checked out from an existing branch
When a worktree is created by checking out an already-existing branch,
the stored baseRefName was set to the branch's own name instead of the
repository's default branch. The diff panel then diffs the branch
against itself, always showing no changes.
Fixes#1873
* fix: fall back to the current branch when no default branch is found
Addresses Greptile review feedback on #1874: resolveRepositoryDefaultBranch
returns null for local-only repos with no origin/HEAD and no local
main/master (e.g. a repo whose trunk is "develop"), which previously
fell back to storing the checked-out branch as its own base — the
exact self-diff bug this PR fixes. Fall back to whichever branch is
currently checked out in the main repo instead.
* fix(server): preserve existing-branch diffs through workspace restore
Carry the resolved comparison base from the worktree creator into the workspace record. Resolve it for occupied branches too, so a fresh suffixed checkout still displays the feature's committed changes.
Co-authored-by: 2takumax-chura <takuma.okayama@churadata.okinawa>
* fix(server): skip deleted default refs when creating worktrees
---------
Co-authored-by: 2takumax-chura <takuma.okayama@churadata.okinawa>
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(app): restore terminal surface after tab hide so CJK cells redraw
Retained panels use display:none. Coming back remasured without treating
that as a surface restore, so Hangul kept 2-cell slots while glyphs painted
at ~1 cell (extra gap). Reload WebGL if the context was lost and force a
refresh when the host grows from zero.
Closes#3434
* fix(server): keep wide-char continuation cells out of terminal snapshots
Wide chars (CJK) occupy two xterm cells, and the second cell reports no
chars. The snapshot capture collapsed that empty continuation to " ", so
every restored Hangul syllable took three columns instead of two. The
injected space was real: selecting and copying the restored row carried
it into the clipboard.
Capture the continuation as "" and have renderTerminalSnapshotToAnsi
skip it. No schema change; snapshots from older daemons still replay
verbatim.
* fix(app): restore retained terminals on presentation
Co-authored-by: seob717 <dev.seob717@gmail.com>
* fix(terminal): limit restoration changes to confirmed CJK corruption
Keep the independently reproduced snapshot fix. Surface recovery behavior remains unverified and belongs with the broader viewport report. Dispose the replay terminal after the regression test.
---------
Co-authored-by: Jason <mildreins@gmail.com>
Co-authored-by: seob717 <dev.seob717@gmail.com>
Co-authored-by: liujin0506 <liujin0506@qq.com>
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(android): declare the camera optional so camera-less devices can install
Requesting CAMERA makes Android treat android.hardware.camera as a required
feature, so Google Play marks devices without a camera as incompatible.
Pairing works without QR scanning, so declare the camera features optional.
Fixes#6303
* test(app): resolve the manifest with the Expo CLI's prebuild packages
Load @expo/prebuild-config and @expo/config-plugins from @expo/cli instead
of the older hoisted copies, and set the F-Droid flag explicitly for each
build profile case.
---------
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* feat(review): edit compact diff comments in a sheet
* feat(review): send diff feedback to workspace agents
Send saved reviews from the diff while retaining file navigation and unrelated composer drafts. Clear acknowledged versions only so comments edited or added during submission remain available.
* fix(review): preserve drafts and coordinate feedback across surfaces
Share pending submission ownership by draft while keeping recipient menus local. Retain editable text across presentation changes and acknowledgement, and reserve safe-area clearance for floating actions.
* fix(review): center a small primary feedback button on desktop
* fix(review): restore focus to feedback button after closing menu
* feat(claude): add Mythos 5.1 with its model capabilities
* feat(claude): gate Mythos on advertised model identity
* fix(claude): retain capabilities for explicitly configured Mythos
* test(usage): load SQLite before timed store tests
The first Windows OMP store case exceeded its five-second budget while subsequent equivalent cases completed under a second. Load the builtin during test collection so its cold initialization does not consume the first test deadline.
* test(workspaces): publish complete service environment fixtures
Waiting for file existence races with writeFileSync creating the file before writing its contents. Rename completed JSON snapshots into place so the existing readiness check cannot expose an empty file.
* feat(server): let clients create internal agents
The daemon's own branch-name and commit-message generators run as
internal agents: never persisted, listed, flagged, announced, or seen by
plugin lifecycle hooks. Nothing outside the daemon could ask for one,
because the wire config schema strips `internal` and the agent.create
transform re-pins it. A plugin that runs a one-turn helper per event
therefore left an archived session in History every time.
Add an optional `internal` flag to `create_agent_request` (inherited by
`agent.create.request`), copy it onto the config in the session's create
path, and skip provider session persistence for it like the generators
do. The client SDK forwards the flag on both creation paths and gates it
once on `server_info.features.internalAgents`.
Two daemon paths assumed every agent is public and had to follow:
`resolveAgentIdentifier` now accepts a live agent by exact id, since the
id set it matched against came from listings that exclude internal
agents and left `wait_for_finish` answering "Agent not found"; and
archiving an internal agent now closes it without writing the record
that `registry.applySnapshot` used to leave behind, so `autoArchive`
does not accumulate hidden files on disk.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(server): keep internal agents whole through failure, archive, and wait
Review follow-ups for internal agents:
- A create that fails after the runtime is registered now closes the
internal agent before rethrowing, so a hidden runtime cannot outlive
the failed request. Public agents keep today's behaviour.
- Archiving an internal agent cascades to attached children like any
parent. cascadeArchiveChildren takes the parent record instead of
reading it back from storage, where an internal agent never is.
- The archived snapshot and last message of an internal agent stay
readable by exact id for ten minutes, so create-then-waitForFinish
returns the finished turn whether the wait lands before or after an
auto-archive. Storage plays that role for public agents.
- The internal-agent e2e scenarios live in the existing
create_agent_request lifecycle suite rather than a new file.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix(server): cascade to live internal children and keep a failed internal create's worktree until it closes
Internal children never reach storage, so the parent archive cascade scans
live agents for them as well. A failed internal create now clears the
created id only after the runtime is archived, so worktree cleanup cannot
delete the directory under a runtime that would not close. Stored agent
payloads carry lastError, so a retained failed helper reports its error
through wait_for_finish instead of the generic message.
* feat(cli,mcp): expose internal agents and document --no-wait
paseo run gains --internal and create_agent gains internal, both forwarded
to the create request. --no-wait becomes the documented spelling for
returning immediately, matching paseo send; --background and --detach stay
as hidden aliases under COMPAT tags.
* test: follow the --no-wait rename and stub getRetiredInternalAgent
* feat: list internal agents on request and add internal workspaces
Internal agents stay hidden by default, but a caller can now ask for
them: `fetch_agents_request` takes `filter.includeInternal`, snapshots
carry `internal: true`, and `paseo ls --internal` shows them in an
INTERNAL column. The app never passes the flag.
Workspaces get the same flag. `workspace.create.request` with
`internal: true` persists a hidden workspace: listings, update
subscriptions, sync reads, empty-project counting, open-by-path, MCP
`list_workspaces`, and plugin workspace hooks all skip it unless the
caller opts in with `filter.includeInternal`. Every agent created inside
an internal workspace is internal and unpersisted; AgentManager applies
that from the workspace id before the plugin hook so creators cannot
opt out. Archive and shutdown enumerate internal agents so nothing
inside an internal workspace outlives it.
Surfaces: `paseo workspace create --internal`, `paseo workspace ls
--internal`, `paseo run --internal --new-workspace`, MCP
`create_workspace({ internal })`, SDK `workspaces.create({ internal })`
and `list({ filter: { includeInternal } })`. Clients gate on
`features.internalWorkspaces`.
Co-Authored-By: Claude <noreply@anthropic.com>
* feat(workspaces): add background visibility with inherited creation defaults
* fix(cli): resolve workspace callers on the selected daemon
* test(workspaces): keep caller integration coverage in server suite
* test(cli): exercise workspace callers through CLI integration harness
* test(mcp): remove obsolete execution flags from browser fixtures
---------
Co-authored-by: marvin-ambrozio <marvin@gustavo.eng.br>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Co-authored-by: Mohamed Boudra <boudra.moha@gmail.com>
* fix(agents): preserve creation when the first prompt fails
A rejected initial prompt must leave a created agent with a turn error instead of returning the user to a draft whose creation key is already committed. Preserve receipt replay without repeating the prompt, and bound fork history to 100,000 characters before it becomes an attachment.
* test(creation): isolate user agent counters from background naming
Workspace naming uses internal provider sessions. Counting and rejecting those sessions made creation assertions depend on installed provider discovery timing. Keep discovery within the test adapters and scope observations to user-created agents.
* test(creation): retry temporary directory cleanup on Windows
Windows CI completed the rejection assertions but failed teardown with EBUSY. Match the daemon test harness bounded removal retries so transient directory locks do not fail creation tests.
* fix(agents): record rejected prompts before their errors
Startup rejection recorded only the system error, leaving the client to append its local prompt after it. Record the rejected submission at the terminal failure boundary so stream and fetched history agree on prompt-before-error ordering.
* fix(agents): require acceptance for delivery notifications
Retain rejected human submissions before their errors without projecting rejected inter-agent prompts as completed deliveries. Reproduces and fixes the notification regression exposed by merge CI.
* test(agents): retain rejected submissions across retries
The existing browser retry test assumed terminally rejected prompts disappeared. Assert the intended prompt-before-error history and a separate accepted retry while preserving draft, attachment, and ambiguous-steering checks.
* test(creation): allow integration latency in raw RPC responses
The peer inherited a one-second polling timeout despite exercising Git and provider startup. Reproduce the timeout with delayed startup and wait within an explicit integration budget while retaining all response and identity assertions.
* feat(agents): show agent messages and completion notifications with sender identity
* fix(agents): preserve human envelope examples and attachment context
* fix(agents): separate message provenance from timeline presentation
Discovery from the home directory walked into ~/Library, where readdir on
TCC-protected group containers blocks forever for a daemon that cannot show
a consent prompt. Each blocked call holds a libuv threadpool thread, so a few
searches wedged every fs operation in the daemon.
- Add an excludedDiscoveryPaths option; discovery skips those paths, while
exact typed paths still resolve and browsing inside lists one level.
- Resolve rooted typed paths directly and anchor the walk under an existing
typed parent instead of walking the whole home directory.
Refs #4879
Anchor the absolutely positioned screen reader install label to its count. Without a containing block inside the carousel, the invisible label escapes the scroll container and expands the document.
* fix(opencode): return browser screenshots to OpenCode v2 agents
The v2 bridge plugin passed a Paseo tool's MCP content straight to
OpenCode. OpenCode v2 tool content is text or a file, so the image block
from browser_screenshot was not valid output: the tool call never
completed and the turn failed.
Convert image blocks to file content with a data URI and keep text
blocks as they are.
Closes#6318
* fix(opencode): show v2 tool images as images, not base64 output
With the bridge now returning screenshots to OpenCode v2 as data URI
file content, the v2 timeline copied each file's URI into the tool
call's output, so the screenshot card showed the whole base64 string.
Keep data URI images out of the tool output, leave an "[image]"
placeholder, and emit each image after the tool call as image markdown,
the way Claude and Codex tool images are shown. History replay uses the
same mapping.
---------
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(schedule): keep firing after a run that could not be recorded
A schedule whose running-run write failed stayed marked as running in
memory, so every later tick skipped it and run-once reported it as
already running until the daemon restarted. The running mark is now
cleared whenever the run ends, including when it could not be recorded,
and the next tick retries the due run.
* test(schedule): skip the unwritable-directory test as root
---------
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(app): catch up provider subagents after the update feed reconnects
Subagent updates are not replayed across a reconnect, so a child that
finished while the app was disconnected stayed "working" until one
subagent was opened and its parent was listed again. Re-list every
parent the app holds once the update feed resubscribes.
Closes#6109
* fix(app): resync a subagent parent even when its first list failed
A parent's subagent track lists the parent once on mount. When that list
failed because the app was disconnected, the parent was never tracked,
so the reconnect resync skipped it and children that later arrived via
live updates stayed stale after the next drop. Track the parent when its
list is requested instead of when it succeeds.
* test(website): expect the plugin directory links now that they are on
1b88f1612 turned PLUGINS_LINKED on, but the e2e test still asserted the
directory was unlinked, failing the typecheck job on main and every PR.
Assert the navigation, footer, and Browse plugins links point at /plugins.
The header check now targets the navigation landmark the homepage renders;
the banner locator matched nothing, so it only passed while asserting zero.
---------
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
Claude Code sends a tool_progress heartbeat every 30 s while a tool is
still running, with that tool's own id as parent_tool_use_id. The Claude
adapter routed any frame with a parent id to the sidechain tracker, so
a main-thread call that ran past 30 s (MCP, WebFetch, WebSearch) became
a "Claude subagent" row with a child action per heartbeat that never
completes, and its card was relabeled Task.
A heartbeat now names no sidechain, so it never reaches subagent
routing.
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(claude): load history from a transcript under another project folder
Claude Code resumes a session by id from any working directory and keeps
writing to the transcript in the project folder where the session started.
When the agent's cwd encodes to a different folder (a moved project, a
migrated daemon, a Windows binary under WSL), the agent kept working but
reopened with an empty timeline. Look the transcript up by session id
across the config dir's project folders when it is not where cwd points.
Refs #4540
* fix(claude): report a failed project folder listing instead of an absent transcript
Take the transcript lookup's inputs as one object, and rethrow read errors other than a missing projects folder so the history load logs them as a failure.
* test(claude): skip the unreadable projects folder case when running as root
---------
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* fix(claude): open autonomous turns only when Claude starts a turn
Paseo guessed from frame shape whether the main session had started work
on its own, and only a result could close the turn that guess opened. A
background subagent's task frames, its own background jobs'
task_notification and rate limit events all arrive with no parent tool
use id, so they looked like main-session work and opened a turn no result
would ever close. The agent then reported "running" while its main
session sat idle, and the app's message queue never drained.
Claude brackets every main-session turn itself: system/init opens it and
exactly one result closes it, whether a prompt, a wake-up after background
work, a slash command or an API error started it. Background subagents
never emit either on the main stream. An autonomous turn now opens only on
that init, and only when no foreground turn is open. The frame heuristic
is removed.
- Paseo launches the CLI with CLAUDE_CODE_EMIT_SESSION_STATE_EVENTS=1,
added as an overlay in buildSdkEnv so the SDK and the usage probe share
one env object. An "idle" session state closes an autonomous turn as a
safety net, never a foreground one. "running" opens nothing: Claude Code
2.1.280 stays "running" for as long as any background subagent is
alive. State frames never reach the timeline.
- An init clears a leftover pendingInterruptAbort. A new turn has begun,
so the interrupted one is over, and a stale flag would swallow the new
turn's result.
Tests that opened autonomous turns from bare frames now send the init
Claude sends. New coverage: helper frames never open a turn while the main
session is idle, a wake-up turn opens on init and closes on its result, a
folded steer keeps one turn while a turn Claude starts after the result
gets its own, an idle closes a turn no result closed, session state never
opens a turn, and the env flag reaches the SDK launch.
docs/agent-lifecycle.md records that only system/init and result bracket
a main-session turn, since the code cannot tell a reader why frame shape
fails.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(claude): interrupt Claude only while its main turn runs, and spare helpers
Every Stop, and every send that replaces a running turn (paseo send, MCP
send_agent_prompt, voice, the app's queue drain, Send now), ends in
query.interrupt(). Without the SDK's perTaskStopAffordance declaration
that call kills every running background subagent, even when the main
session is idle and there is nothing of ours to stop.
- Declare perTaskStopAffordance, so an interrupt aborts only the main
turn, like Esc in Claude Code. Claude stops a helper itself with its
TaskStop tool.
- Track whether a main-session turn is in flight from Claude's own
brackets: set by system/init and by a pushed message's
command_lifecycle "started" (/compact works for seconds before its
init), cleared by the result or an idle session state.
- Call query.interrupt() and set pendingInterruptAbort only when that
turn is in flight, or the cancelled turn's own message has already been
started. Otherwise withdraw our not-yet-started messages with
cancel_async_message instead: an idle interrupt reaches only helpers,
and with no result to follow it the flag would linger.
- Track the foreground prompt with the queued steers until Claude starts
it, so Stop withdraws a message still waiting in the queue. A turn
stopped while the Claude process is still starting never sends it.
- Accept command_lifecycle "cancelled" as terminal, so a cancelled steer
no longer waits to be withdrawn or keeps denying permission cards.
Interrupt tests now wait for the pump to read the init before stopping,
since that is when Paseo learns a turn is in flight. New coverage: a stop
before Claude starts the message withdraws it without interrupting, a
stop while only helpers run never reaches Claude, a stop mid-turn
interrupts once and drops the aborted result, a new turn clears a flag
no result will consume, the SDK option is passed, and a cancelled steer
is not withdrawn again.
docs/providers.md described the old unconditional interrupt; it now states
when the Claude adapter interrupts and why it declares the affordance.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(claude): keep a helper spawned in the background out of a turn's cancel
When a turn is canceled, Paseo terminalizes the subagents that were
running in its foreground and skips the backgrounded ones, which outlive
it. It only learned a child was backgrounded from task_updated's
patch.is_backgrounded, which Claude sends when a foreground task moves to
the background later. A child spawned in the background carries the flag
on task_started instead, and a resumed subagent always does, so the
track marked those helpers "canceled" while they kept working.
Read is_backgrounded from task_started as well. docs/agent-lifecycle.md
said task_updated was the only signal; it now names both.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(app): drain queued messages as steers
The composer queue drains when an agent goes idle. It sent each message
with no activeTurnBehavior, which the daemon treats as "interrupt": if
the agent had just started a turn of its own (a Claude wake-up after
background work finished), the drained message stopped that turn and
replaced it.
Send drained messages with activeTurnBehavior "steer". A drain that races
such a turn folds into it; with no turn running the daemon starts a
normal one, as before. "steer" is an existing protocol value, so older
daemons need nothing new.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(claude): cover background helpers against real Claude
Every other test in this file tells Claude "Do not use a background
task", so a message sent or a Stop pressed while a backgrounded subagent
runs was never exercised. Four real-Claude scenarios cover it:
- "a background subagent keeps running after a message is sent while it
works": one helper, the main turn ends, the parent is not reported
running, a message in that window is answered, and the helper is never
read as canceled and completes.
- "a helper's own nested background job does not phantom-run the parent
or kill the helper": a helper starts its own background Bash job, then
keeps working in the foreground. The parent stays idle across the
nested job's completion (sampled for 30 s around a 5 s job), a message
sent in that window is answered, and the helper reports CHILD_DONE.
- "the app's default steer survives a backgrounded helper while the main
session works": the app's default send behavior (steer) does not kill
a helper when it interrupts the main session's own turn.
- "Stop cancels the main session's own turn but spares its background
helper": the same for Stop.
The wake-up turn Claude starts after a helper finishes lands a moment
after the helper reads completed (0.1 to 5 s in recorded traces), and the
daemon announces it as a pushed "running" agent_update, not as a
turn_started stream event. The checks poll for it and read the final idle
once the wake-up settles. waitForFinish can return a stale "idle" right
after a send, and the push stream can miss a message's final text when it
resolves, so the tests read fetchAgentTimeline like the rest of the file.
Helper status comes from agent.provider_subagents.list, since agent_stream
carries no provider_subagent events.
The scenarios need the previous commits together. Run against the frame
heuristic this branch removes, the nested-job test fails: the parent
flips to running once the nested job finishes and stays there.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(claude): interrupt a stopped prompt Claude has already taken
Stop before Claude announces a prompt started withdraws it with
cancel_async_message instead of interrupting. Claude can dequeue a
message for its next turn before announcing it, and withdrawing it is
then a no-op that reports cancelled: false, so the stopped prompt kept
running. Interrupt whenever a message could not be withdrawn.
Co-authored-by: Isak Persson <isak.a.persson@gmail.com>
* fix(claude): withdraw a stopped prompt only once
When the withdrawal of a stopped prompt failed, the interrupt path
withdrew a second time, which could cancel a prompt sent while the first
withdrawal was in flight. Withdraw once and interrupt on failure.
The real-Claude steer scenario now waits for the background helper to
exist before steering, so it proves the helper survives the steer.
Co-authored-by: Isak Persson <isak.a.persson@gmail.com>
---------
Co-authored-by: Isak Persson <isak.a.persson@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
* test(website): reproduce plugin search dropping a term typed before hydration
Typing in the plugin search box before the page's scripts run leaves the
term in the box but not in React state, so Enter searches without it.
* fix(website): keep a plugin search term typed before the page loads
Text typed into the plugin search box before hydration stayed in the box
but not in component state, so Enter searched without it and the clear
button never appeared. Adopt the box's value once the page hydrates.
Closes#6286
* test(website): cover live browse search typed before the page loads
---------
Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com>
paseo-bot[bot]