Desktop users can now disable thinking for new sessions, and Anthropic-compatible providers that opt in receive explicit disabled-thinking requests across main turns, side queries, and AI title generation.
Constraint: DeepSeek/Kimi/GLM Anthropic-compatible endpoints need a non-thinking path without scattering provider-specific logic through the CLI.
Rejected: Per-model if/else branches | centralized provider preset env keeps the native CLI surface smaller and easier to audit.
Confidence: high
Scope-risk: moderate
Tested: bun test src/utils/__tests__/thinking.test.ts src/server/__tests__/title-service.test.ts src/server/__tests__/provider-presets.test.ts src/server/__tests__/conversations.test.ts
Tested: cd desktop && bun run test -- generalSettings.test.tsx
Tested: cd desktop && bun run lint
Tested: cd desktop && bun run build
Tested: agent-browser DeepSeek desktop E2E with transparent proxy captured main and title requests with thinking.type=disabled
The session inspector now asks the resumed CLI for a fast structural context estimate instead of forcing the full token-counting API path. This preserves live CLI state for system prompt, tools, MCP tools, skills, and messages while avoiding the 20s timeout that made historical sessions appear stuck or fall back to transcript-only estimates.
Constraint: Inspector requests must return quickly for third-party providers and historical resumed sessions.
Rejected: Increase the server timeout | the slow path can still block on provider token counting and keeps the UI feeling broken.
Confidence: high
Scope-risk: narrow
Directive: Keep interactive inspector context on the estimateOnly control path unless the UI explicitly supports a slow precise refresh.
Tested: bun test src/server/__tests__/conversations.test.ts -t 'structured session inspection|Sonnet 4.6 transcript usage'
Tested: cd desktop && bun run test -- --run src/i18n/index.test.tsx
Tested: cd desktop && bun run build
Tested: direct /api/sessions/:id/inspection?includeContext=1 returned live context in 0.046867s with System prompt, System tools, MCP tools, Messages
Tested: agent-browser automation verified no loading/error and visible System prompt/System tools/MCP tools/Messages
Not-tested: bare root tsc --noEmit, because current tsconfig scans existing desktop/src-tauri/target generated binary assets unrelated to this change
Default desktop sessions with an active third-party provider were only passing the selected model into the CLI startup path. That allowed the child CLI to fall back to stale cc-haha settings env instead of the provider index, so the UI-selected provider could diverge from the key and endpoint used by the subprocess.
Route default sessions through the active provider id and make third-party Sonnet capability overrides explicit with a non-empty sentinel. This keeps provider runtime env sourced from providers.json and avoids treating third-party Sonnet models as first-party thinking-capable models.
Constraint: Third-party provider behavior is the product baseline for desktop users.
Rejected: Rely on synced cc-haha/settings.json env | it can be stale and caused provider/key routing mismatches.
Confidence: high
Scope-risk: narrow
Directive: Third-party desktop sessions must carry providerId into ConversationService; do not depend on global settings env for active providers.
Tested: bun test src/server/__tests__/conversations.test.ts src/server/__tests__/conversation-service.test.ts src/server/__tests__/provider-presets.test.ts src/utils/__tests__/thinking.test.ts
Tested: git diff --check
Third-party Anthropic-compatible providers can expose Claude model names without supporting every first-party runtime capability. Desktop sessions now preserve provider-specific capability overrides, keep selected provider runtime state across reconnects and restarts, and keep the provider dialog behavior aligned with the updated presets.
Constraint: Custom ANTHROPIC_BASE_URL endpoints must not be treated as first-party Anthropic for adaptive thinking support.
Rejected: Rename provider default Sonnet models | that would hide the compatibility issue instead of fixing runtime capability detection.
Confidence: high
Scope-risk: moderate
Directive: Do not enable Claude first-party capability defaults for custom Anthropic-compatible base URLs without provider verification.
Tested: bun test src/utils/__tests__/thinking.test.ts src/server/__tests__/conversation-service.test.ts src/server/__tests__/provider-presets.test.ts src/server/__tests__/conversations.test.ts
Tested: cd desktop && bun run lint
Tested: cd desktop && bun run test src/components/shared/Modal.test.tsx --run
Tested: git diff --check
Desktop sessions can switch provider and model while a CLI subprocess is already alive, so the server now serializes runtime restarts and marks provider-managed launches to prevent stale settings env from overriding the selected provider. Provider settings also write API key env consistently and clear stale managed keys before syncing.
This includes the related desktop/docs brand asset refresh and keeps the desktop locale default in Chinese, with tests updated to match the current provider semantics.
Constraint: Session-scoped model selection must win over cc-haha/settings.json and inherited ANTHROPIC_* values.
Rejected: Store the selected model as a global provider activeModel | chat runtime selection is per session.
Confidence: high
Scope-risk: moderate
Directive: Do not remove CLAUDE_CODE_PROVIDER_MANAGED_BY_HOST without validating Desktop provider switching against stale settings env.
Tested: bun test src/server/__tests__/conversation-service.test.ts src/server/__tests__/conversations.test.ts src/server/__tests__/providers.test.ts src/server/__tests__/providers-real.test.ts
Tested: cd desktop && bun run test src/stores/settingsStore.test.ts
Tested: cd desktop && bun run lint
Tested: git diff --check
Not-tested: Full desktop production package/signing.
The desktop app already shipped a bundled sidecar, but only desktop-managed
sessions could see it. This change installs a `claude-haha` launcher into the
user bin directory, wires PATH setup so new terminals can resolve it, and keeps
desktop installer sessions aligned on the same bundled sidecar resolution path.
The desktop install surface now also reports whether the launcher is ready or
still waiting on a terminal restart.
Constraint: The worktree already contains unrelated icon, docs, and UI changes, so this commit stages only the bundled CLI launcher slice
Rejected: Tell users to install the official Claude CLI separately | it breaks the desktop out-of-box install story
Rejected: Keep the bundled CLI reachable only inside desktop-managed shells | system terminals would still be unable to call the packaged runtime
Rejected: Symlink directly into the app bundle instead of copying to user bin | moving or replacing the app bundle would leave a stale launcher behind
Confidence: high
Scope-risk: moderate
Reversibility: clean
Directive: Proxy-backed non-Anthropic providers still depend on the desktop server; do not assume this launcher makes every provider fully standalone
Tested: bun test src/server/__tests__/desktop-cli-launcher.test.ts src/server/__tests__/settings.test.ts; bun test src/server/__tests__/conversation-service.test.ts; bun test src/utils/shell/bashProvider.test.ts; cd desktop && bun x vitest run sidecars/launcherRouting.test.ts src/components/settings/InstallCenter.test.tsx; cd desktop && bun run lint; cd desktop && bun run build; cargo check --manifest-path desktop/src-tauri/Cargo.toml
Not-tested: Manual packaged desktop app install plus real Terminal/iTerm/PowerShell invocation on fresh macOS and Windows machines
The desktop app could read plugin-produced skills and agents, but it had no
plugin control plane of its own. This adds a dedicated Settings tab backed by
server-side plugin APIs so installed plugins can be inspected, enabled,
disabled, updated, reloaded, and uninstalled from the WebUI.
The implementation also teaches browser-based desktop dev sessions to honor a
custom backend URL, which made it possible to run isolated worktree ports for
real UI automation. During verification, the long-lived desktop server kept a
stale installed-plugin snapshot after external CLI mutations, so cache clearing
now resets that session-level plugin installation state as well.
Constraint: Desktop WebUI needed an isolated backend URL instead of the hard-coded 127.0.0.1:3456 fallback
Constraint: Reuse existing plugin operations and loaders instead of rebuilding plugin lifecycle logic in the desktop layer
Rejected: Fold plugin management into Skills or Adapters | mixed unrelated lifecycles and hid plugin-specific health/actions
Rejected: Expose only read-only plugin status in desktop | did not satisfy enable-disable-reload-uninstall verification needs
Confidence: high
Scope-risk: moderate
Reversibility: clean
Directive: Keep desktop plugin actions routed through the shared plugin operation layer and clear installed-plugin session caches when plugin state changes externally
Tested: cd desktop && bun run lint
Tested: cd desktop && bun run test -- src/__tests__/pluginsSettings.test.tsx
Tested: bun test src/server/__tests__/plugins.test.ts src/server/__tests__/skills.test.ts
Tested: Browser automation against isolated ports 15120/38456 covering discord plugin list/detail/disable/apply/enable/update/uninstall flows
Not-tested: Full desktop session runtime parity with CLI /reload-plugins AppState refresh beyond the new desktop API path
Desktop sessions were missing a visible request_access approval path and could
mis-detect their own app window as an unapproved frontmost target, which caused
Computer Use clicks to fail even after opening the intended app. On macOS, text
entry was also split across inconsistent clipboard and keystroke paths, making
Electron inputs unreliable for Chinese and short strings.
This change adds a desktop approval bridge over the existing session websocket,
renders a dedicated desktop approval modal, threads the real desktop bundle id
into the Computer Use executor, and switches macOS clipboard typing onto the
native pasteboard plus system paste shortcut path. It also makes tool error
results expandable in the desktop chat UI so frontmost-gate failures are fully
visible during debugging.
Constraint: Desktop sessions run the CLI over the SDK websocket path, so Ink tool JSX dialogs are not visible there
Constraint: macOS IME and Electron text inputs are unreliable with pyautogui.write and generic hotkey synthesis
Rejected: Reuse CLI setToolJSX dialogs in desktop mode | no transport for mid-call Ink UI over the SDK bridge
Rejected: Keep shell pbcopy/pbpaste for clipboard typing | inconsistent with NSPasteboard path and less reliable for Chinese text
Confidence: medium
Scope-risk: moderate
Reversibility: clean
Directive: Keep desktop Computer Use approvals and macOS text-entry behavior on a single bridge/path; avoid reintroducing separate CLI-only and desktop-only codepaths for the same action
Tested: python3 -m unittest runtime/test_helpers.py
Tested: bun test src/utils/computerUse/permissions.test.ts src/server/__tests__/conversation-service.test.ts
Tested: cd desktop && bun run test ComputerUsePermissionModal chatStore
Tested: cd desktop && bun run test chatBlocks
Tested: cd desktop && bun run lint
Not-tested: End-to-end manual Computer Use interaction against a live Electron target app on macOS
Windows desktop sessions were failing during early hook execution because the
runtime assumed Git Bash was always installed. That made a missing Git
installation look like a fatal CLI startup regression even though PowerShell is
usually available and sufficient for default hook execution.
This change turns Git Bash lookup into a best-effort probe, leaves SHELL unset
when Git Bash is missing, and falls back to PowerShell for default Windows hook
execution. Explicit bash hooks still fail with a direct installation hint, but
they no longer hard-exit the whole process.
Constraint: Windows users may have PowerShell but no Git for Windows installed
Rejected: Keep hard-exiting on missing Git Bash | makes first-run desktop chat fail for a recoverable missing dependency
Rejected: Bundle Git for Windows into the app | larger distribution and higher maintenance burden than a runtime fallback
Confidence: medium
Scope-risk: moderate
Reversibility: clean
Directive: Treat Git Bash as an optional Windows capability unless a code path explicitly requires bash semantics
Tested: bun test src/server/__tests__/conversation-service.test.ts
Tested: bun -e "await import('./src/utils/windowsPaths.ts'); await import('./src/utils/hooks.ts'); console.log('module-import-ok')"
Not-tested: End-to-end desktop chat startup on a clean Windows Server host without Git installed
Not-tested: Explicit bash hook execution on Windows after the fallback path
Related: GitHub issue #62
This bundles the pending desktop/server team-session fixes with the local adapter recovery changes already in the worktree. The team path now keeps teammate membership stable under concurrent spawns, surfaces real teammate identities in the desktop UI, and allows direct interaction with member transcripts. The adapter changes recover automatically when stale thinking signatures invalidate an existing session.
Constraint: Team config writes can happen concurrently while multiple reviewers spawn in parallel
Constraint: Desktop member views must follow mailbox/transcript semantics rather than hijacking teammate runtime sessions
Rejected: Keep relying on config.json alone for member discovery | in-process teammates can be lost after concurrent writes
Rejected: Open teammate sessionIds as normal desktop sessions | would attach a second CLI instead of the running teammate
Confidence: medium
Scope-risk: moderate
Reversibility: clean
Directive: Preserve locked team-file mutation for any future teammate registration path and keep teammate labels sourced from member names before agent types
Tested: bun test src/server/__tests__/teams.test.ts src/server/__tests__/team-watcher.test.ts
Tested: cd desktop && bun run test --run src/stores/chatStore.test.ts src/pages/ActiveSession.test.tsx
Tested: cd desktop && bun run lint
Tested: cd desktop && bun run build
Not-tested: Manual end-to-end validation against a live Agent Teams run in the desktop app
Cuts the macOS .app from 435MB → 152MB (-65%) and the DMG from 113MB → 60MB
(-47%) by inlining src/server and src/entrypoints/cli into the bun-compiled
sidecar binaries instead of dynamic-importing them from disk at runtime.
Architectural change
====================
Before:
desktop/sidecars/server-launcher.ts → bun build --compile (≈57MB shell)
└─ at runtime: dynamic file:// import of <appRoot>/src/server/index.ts
which transitively requires ALL of src/ + the entire root node_modules/
to be shipped as Resource. tauri.conf.json copied 254M of node_modules
and 47M of src/ into Contents/Resources/app/ on every build.
After:
desktop/sidecars/server-launcher.ts → bun build --compile (≈65MB)
└─ uses `await import('../../src/server/index.ts')` with a literal
specifier so bun's bundler walks the whole graph statically and
inlines everything into the binary.
Same treatment for cli-launcher.ts → src/entrypoints/cli.tsx.
Resolver gymnastics
===================
This fork carries dozens of ant-internal feature() gated require/import
calls referencing modules that simply don't exist on disk
(cachedMicrocompact, devtools, proactive, coordinator, etc). Bun's resolver
walks the static dep graph BEFORE bun:bundle macro DCE, so even though
the dead branches never execute at runtime, they still fail to resolve
at compile time.
Two complementary mechanisms:
1. desktop/scripts/scan-missing-imports.ts walks src/, regex-greps every
relative import / require / type-import specifier, and writes a Proxy
noop stub for any target that doesn't exist on disk. Stubs are tagged
with "@generated stub from scan-missing-imports" for idempotency. Text
resources (.md / .txt / .json) get appropriate format-specific stubs.
Runs as a pre-step inside build:sidecars.
2. desktop/scripts/build-sidecars.ts adds an `external: [...]` list for
bare-package optional deps not in package.json (OTLP exporters,
@aws-sdk/*, @anthropic-ai/{bedrock,vertex,foundry,mcpb}-sdk,
@azure/identity, fflate, turndown, sharp, react-devtools-core).
These remain runtime imports, fail benignly when their gating env
var or feature flag is off.
Tauri side
==========
- desktop/src-tauri/tauri.conf.json: dropped all `resources` entries.
Was 7 entries totaling ≈301MB. Now `{}`.
- desktop/src-tauri/src/lib.rs `resolve_app_root` no longer calls
BaseDirectory::Resource (the app/ resource dir doesn't exist anymore);
instead returns the directory of the current sidecar exe. The launchers
still accept --app-root for backward compat with conversationService's
CLI subprocess spawn.
Optimisations
=============
- bun build now uses minify whitespace+identifiers+syntax. Saved another
≈16MB across both binaries (server: 72MB→65MB, cli: 75MB→66MB).
Bonus fix
=========
src/services/remoteManagedSettings/index.ts had a typo importing
'./securityCheck.jsx' instead of '.js'. Bun's runtime resolver tolerated
it; bun build didn't.
Verification
============
- Both binaries boot successfully in /tmp with no src/ or node_modules/
on disk. Verified `claude-cli --version` returns the build version,
`claude-cli --help` prints the full Commander spec, and claude-server
starts CronScheduler + listens on the requested port.
- bun test on src/ shows 358 pass / 45 fail / 2 errors vs main baseline
of 359 / 44 / 2 — net 0 new failures (1 different flake direction).
All 44 baseline failures pre-exist on main and are unrelated.
- Full DMG round-trip via build-macos-arm64.sh succeeds; new bundle
installs cleanly in /Volumes/.
Bundle size summary
===================
metric baseline after P0 delta
Resources/app/ 301 MB 0 MB -301 MB
MacOS/claude-server 57 MB 65 MB +8 MB
MacOS/claude-cli 57 MB 66 MB +9 MB
MacOS/claude-code-desktop 18 MB 18 MB —
─────────────────────────────────────────────
.app total 435 MB 152 MB -283 MB (-65%)
.dmg 113 MB 60 MB -53 MB (-47%)
Generated stub files (173 of them under src/) are committed for
reproducibility — the scanner is idempotent and will re-create them
identically on every build, but tracking them avoids dirty working trees
on first compile.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Bypass all three gating layers (compile-time feature flag, subscription
check, GrowthBook remote config) and replace Anthropic's private native
modules (@ant/computer-use-swift, @ant/computer-use-input) with a Python
bridge using pyautogui + mss + pyobjc. Works on any macOS with any
Anthropic-protocol compatible model.
The /buddy command was completely disabled by bun:bundle feature('BUDDY')
flag which evaluates to false at runtime. Removed all feature('BUDDY')
checks across the codebase to register the command, and added keyboard
event handling (q/Enter to dismiss) which was missing from the UI.
