⌘H is reserved by macOS for "Hide Application" and pre-empts app-level
menu bindings before they ever fire, so the Command History shortcut
never actually worked. Moved to ⇧⌘H and updated all references (in-app
help, macOS Help Book, README).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Adds three new Help sections covering the 2.4.1 features: connecting
external stdio MCP servers, Calendar/Reminders/Location access with
its approval flow, and parallel read-only research sub-agents. Also
notes French as a supported language and that OpenRouter's dedicated
image models are merged into the model picker automatically. Cleans
up a stray misplaced HTML comment above the Anytype section.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Two contributing layout issues in the chat message bubble:
1. MessageRow's content VStack (icon + text HStack) had no
.frame(maxWidth: .infinity), so it sized to its content's ideal
width instead of the space actually available.
2. swift-markdown-ui renders paragraphs with mixed inline styling
(bold/italic runs next to plain text) as concatenated Text(+)
segments, which on macOS report their ideal unwrapped single-line
size for height purposes instead of wrapping — truncating with "…"
regardless of window width. Plain single-style paragraphs (a single
Text) weren't affected, which is why some lines wrapped fine and
others didn't.
Fixed by adding .frame(maxWidth: .infinity, alignment: .leading) to
the MessageRow content stack, and .fixedSize(horizontal: false,
vertical: true) to the markdown paragraph label so height is
recomputed for the width actually given instead of the ideal width.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The tool loop's max-iterations and empty-response fallbacks were showing
placeholder assistant bubbles ("[Tool loop reached maximum iterations]",
"[No response from the model — retrying]") followed by a "↩ Continuing…"
system message before silently re-running. None of that added anything
for the user, so the auto-continue now happens without any visible
message when there's no real content to show; genuine partial content
is still displayed as before, and usage/cost tracking is unaffected.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Lets the AI connect to any external stdio MCP server (e.g. safaridriver
--mcp) configured in Settings, with tools auto-discovered and prefixed
by server slug. Includes crash detection with backoff restart (5s/15s/30s)
and a Settings UI to add/enable/disable/remove servers.
Fixes the temp-dir allowlist in MCPService.isPathAllowed to also match
/tmp and /private/tmp (not just NSTemporaryDirectory(), which resolves
to a different per-user Darwin temp dir) so the MCP file tools can
actually read files external servers and image generation write there.
Also switches the Add Server sheet's argument parsing to a quote-aware
tokenizer so args containing spaces survive intact.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Fetches /api/v1/images/models in parallel with /models and merges results
into the model picker. Image-only models (e.g. Sourceful, Seedream, Flux
via this endpoint) were previously invisible since they don't appear in the
standard /models endpoint.
Models from the images API get usesImagesAPI=true and route through a new
generateImageAPIResponse() path in ChatViewModel that POSTs to /api/v1/images
with {model, prompt} instead of the chat completions endpoint. The response's
b64_json data is decoded and displayed via the existing GeneratedImagesView.
Cost is taken directly from the usage.cost field in the images API response
(USD per image) via a new rawCostUSD field on ChatResponse.Usage, bypassing
the token-based calculateCost() path used for chat models.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Personal Data Tools: native Calendar, Reminders, Contacts (hidden pending
Apple TCC fix in beta), and Location & Maps access via EventKit, Contacts
framework, and MapKit. Write actions (create event/reminder, complete
reminder) gate through an approval sheet. Four hardened-runtime entitlements
added to oAI.entitlements; Info.plist usage strings added for all services.
Personal Data section shows a β badge while Contacts is hidden.
2nd Brain always-trust: inline toggle on the Agent Skills row for the skill
named "2nd Brain" skips the bash approval dialog when the command contains
.brain_helper.py, gated by three runtime checks in MCPService.
Research agents: spawn_research_agents tool runs up to 5 concurrent read-only
sub-agents (read_file, list_directory, search_files, web_search — no write,
no bash, no nesting). Bounded by maxConcurrentAgents setting (default 3) and
a hard ceiling of 8 tasks. Added items field to Tool.Function.Parameters.Property
for JSON Schema array support; wired into AnthropicProvider.convertParametersToDict.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Document Jarvis integration, Combine Conversations, model category
filter, sidebar navigation, prompt caching, and the 2nd Brain trust
toggle. Clarify that localization is AI/machine-translated rather than
reviewed by native speakers. Normalize all em-dashes to plain hyphens.
French (fr) added as a 5th supported language; full catalog translated.
Also caught up nb/sv/da/de for ~300 strings added since the last
localization pass (Jarvis, Anytype, model categories, reasoning effort,
Combine Conversations) plus Button/Toggle/Menu/CommandMenu titles and
custom sectionHeader/row helpers in Settings that were never extracted
by prior tooling, leaving Settings and the View menu English-only
regardless of locale.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
rune