* feat(agents): add first-class Muse Code harness
Add Muse as a supervised Orca agent across desktop, mobile, session history, source control, local hooks, SSH, WSL, and native Windows. Preserve user settings, support Muse 1.3 hook environment allowlists, and recognize versioned foreground processes. Include question, waiting, completion, resume, and readiness coverage.
Co-authored-by: homesh-dev <300847526+homesh-dev@users.noreply.github.com>
Co-authored-by: jeffhuen <32542276+jeffhuen@users.noreply.github.com>
Co-authored-by: John Cusack <johncusackccm@gmail.com>
Co-authored-by: Adrien De oliveira <75085839+adriendeoliveira@users.noreply.github.com>
* test(agents): cover Muse remote hook registration
* test(agents): cover Muse hook and source-control contracts
* test(agents): exclude Muse hook metadata from script mode check
* test(agents): keep Muse skill picker coverage stable
* test(ai-vault): include Muse in every-agent fixture
* test(mobile): repin Muse agent icon closure
* fix(muse): detect questions and approvals from structured Muse signals
Muse 1.3 fires no hook for request_user_input, so a pending question left
the pane "working". Its internal reminder subagents also post hooks with
their own session ids (even after Stop), which surfaced "tool failed" rows
and flipped finished panes back to working.
- Read pending questions from Muse's session log
(user_input_prompt_requested/settled) via the existing transcript poll,
now generalized from Codex subagents to Muse on main and relay.
- Drop child-session hooks (SubagentStart ids, or turn_id === session_id).
- Treat Notification permission_prompt as the approval wait; PermissionRequest
also fires for auto-approved calls, so it only caches the approval card.
- Ignore Notification copy as the prompt; poll replays are not new prompts
or turn boundaries.
- Allowlist USERPROFILE so Windows cmd AutoRun doesn't fail every hook.
* perf(muse): parse only question events from the session log
Most Muse session-log lines are large model/tool records. Filter raw lines
by the user_input_prompt_ marker before JSON.parse via an optional
readJsonlCursor line filter.
* fix(muse): unwrap batched log records and scope questions to the live turn
Review follow-ups: question events inside retained_frame batches were
skipped, and a question left open by a crash or interrupt stayed pending
for the pane's life. Share the history scanner's retained_frame unwrapper,
and only report a pending question whose run_id matches the hook turn_id.
* refactor(muse): drop type assertion in retained_frame unwrap
* fix(agent-hooks): satisfy exhaustive-switch lint in transcript poll policy
---------
Co-authored-by: Adrien De oliveira <75085839+adriendeoliveira@users.noreply.github.com>
* fix(commit-message): keep Windows paths intact in agent command overrides
`tokenizeCustomCommandTemplate` applies POSIX backslash-escape rules on every
platform. On Windows `\` is the path separator, so a native absolute path in an
agent command override is silently destroyed:
C:\Windows\System32\WindowsPowerShell\v1.0\powershell.exe
-> C:WindowsSystem32WindowsPowerShellv1.0powershell.exe
which is then reported as not found on PATH. The agent *startup* path already
routes Windows shells to the Windows tokenizer, but the commit-message AI path
still calls the generic tokenizer directly, so overrides, extra CLI args and
custom commands there are all affected.
The tokenizer gains an explicit `'escape' | 'literal'` mode rather than reading
`process.platform`, because the same template can be parsed on one host and
executed on another. `'escape'` stays the default, so POSIX behaviour — where
`foo\ bar` is deliberately one token — is unchanged.
`'literal'` is selected only where the command provably runs on native Windows:
a LOCAL target, on win32, with no WSL distro. A WSL target runs a Linux binary
inside the distro, and a remote target runs on a host whose platform this
process cannot see; both keep POSIX escaping.
Fixes#11375
* test: pin the platform decision for literal-backslash parsing
commandBackslashMode is the only place that reads the platform, so it is where
this can be wrong in the direction that matters — applying Windows rules to a
command that will actually run under a POSIX shell. WSL and remote targets are
pinned explicitly; both were previously untested.
* fix(source-control-ai): stop duplicating singleton CLI flags in agent argv
Recipe CLI arguments and agent command overrides were appended on top of
Orca's generated flags, so a user-supplied --model produced a repeated
flag. yargs collapses a repeated flag into an array, crashing OpenCode
with "j.split is not a function"; clap rejects it outright for Codex.
Declare the at-most-once option groups per agent spec and fold every
user-supplied occurrence into the generated slot, with recipe args
outranking a command-override prefix.
Fixes#12305
* fix(source-control-ai): preserve command override argv order
* WIP: Changes before auto-review fixes
* Customize Source Control AI action recipes
- Add per-action CLI arguments for generation and launch flows so saved
recipes can select model flags without putting prompts in argv.
- Share the text-generation dialog between commit messages and hosted-review
details, with first-run defaults and repo/global recipe support.
- Launch fix-check agents directly from saved recipes and harden prompt/template
handling for invalid args, blank prompts, and inherited variables.
* Allow custom commands for source control action recipes
- Let text action recipes save and resolve the custom-command sentinel
- Add settings UI for custom command recipes and preserve per-action defaults
- Split large source-control dialogs and direct launch helpers into focused modules
- Keep launch actions from treating custom text agents as runnable TUI agents
* Add per-repo Source Control AI enablement, custom command, and save-targ
- Repo overrides now support `enabled` and `customAgentCommand`, letting
repositories opt in/out of Source Control AI independently and supply a
repo-scoped custom command that takes precedence over the global one.
- Recipe-save dialogs gained a save-target selector ("Don't save / Save for
this repo / Save as global default") replacing the old boolean checkbox,
routing saves through the new `saveSourceControlActionRecipe` helper in
`source-control-ai-recipe-save.ts`.
- `normalizeRepoSourceControlAiOverrides` now returns `undefined` for empty
objects and passes the `null` sentinel through the IPC/RPC layer so the
persistence layer can clear repo overrides cleanly.
- `resolveSourceControlLaunchPlatform` resolves the correct shell platform
for SSH and WSL worktrees so agent launch commands are built correctly.
- Settings UI gained `RepositorySourceControlAiEnablement` and
`RepositorySourceControlAiCustomCommand` rows; draft/label logic was
extracted into focused modules to stay within lint line limits.
* Extract action recipe defaults into own component and use id-prefixed wo
- Move action recipe draft state and UI out of CommitMessageAiPane into SourceControlAiActionRecipeDefaults and source-control-ai-action-recipe-draft.ts to respect the max-lines lint rule
- Use toRuntimeWorktreeSelector() across all runtime git RPC calls so the runtime can resolve worktrees by ID rather than path
- Fix SSH launch platform resolution to use the repo's connection when the newly created worktree isn't hydrated yet
- Add edit and delete handlers for PR conversation comments with confirmation dialog
- Use text-status-success design token instead of hardcoded text-emerald-500
* add more search keyword
* Rename "Enable Source Control AI defaults" to "Show Source Control AI ac
* fix test
* Remove unused imports and variable assignment in launch-work-item-direct
* Fix test mocks to use `mocks.store` instead of `storeState.value` for di
* Extract Source Control AI logic into focused modules with fix-checks dia
---------
Co-authored-by: Orca <help@stably.ai>
NeilandAdrien De oliveira
Brennan Benson
Jinjing
Jinwoo HongandOrca
