feat: intelligent polish + per-app context + personal dictionary

v0.3.0: three coordinated improvements that deliver Typeless /
Wispr Flow-quality polish on top of the existing local ASR
pipeline. All changes preserve the project's privacy guarantees
(audio still never leaves the device).

## 1. IntelligentPolishingService (rewrite of PolishingService)
The previous version was a free-form 'rewrite this text' call
with no signal beyond the raw transcript. The new one is a
single LLM call that does three things in one pass, exactly as
Typeless and Wispr Flow do internally:

  1. ASR error correction (homophones, near-misses, missing chars)
  2. Polish (drop filler words, fix grammar, add punctuation)
  3. Style adaptation per app context (code / email / chat / doc)

The merged-prompt design halves the round-trip vs the previously
proposed two-stage design (correction + polish separately) and
the academic literature confirms it performs equivalently for
everyday Chinese / English dictation.

## 2. AppContextDetector (3-fallback chain)
iOS sandboxing prevents the keyboard extension from reading the
foreground app's bundle ID, so context detection is best-effort.
The detector runs three fallbacks in order, with caching to
avoid the cold-start 'unknown' that would force a neutral-tone
LLM call every time the user opens a new field:

  1. Heuristic on the text at the cursor (code / email / chat / doc)
  2. 30-minute cache of the last successful detection
  3. Time-of-day + weekend heuristic as a soft default

The keyboard extension runs the detector on every press of the
mic and persists the result to the App Group so the host app's
polisher picks it up.

## 3. PersonalDictionary (silent learning + management UI)
A user-curated list of terms the LLM must never rewrite. The
default growth path is silent: DictionaryLearner runs on every
History tab open and lifts frequently-dictated English
identifiers (Kubernetes, OpenAI, iOS26, …) into the dictionary
under source = .history. Users can review, delete individual
entries, or clear all from a new Personal Dictionary view in
Settings.

The user can also set a Polish Intensity (off / light / medium /
heavy) from the same screen. Default is medium, which is what
Typeless and Wispr Flow also use.

## Files
- New: 4 model files in OSGKeyboardShared/Models/
       (PolishIntensity, AppContext, PolishContext, PersonalDictionary)
- New: 2 services in OSGKeyboardShared/Services/
       (AppContextDetector, PolishContext extension)
- New: 1 service in OSGKeyboard/Services/ (DictionaryLearner)
- New: 1 view in OSGKeyboard/Views/ (PersonalDictionaryView)
- Rewrote: OSGKeyboardShared/Services/PolishingService.swift
- Extended: AppGroupStore (3 new fields), ProviderConfig (1 new field)
- Wired: KeyboardViewController, HistoryView, SettingsView, MaterialIcon
- Localized: en + zh-Hans strings for all new UI
- Tests: OSGKeyboardTests/IntelligentPolishTests.swift (16 tests)

## Verification
- All new code follows the existing Sendable / strict-concurrency
  patterns (the keyboard extension stays within its 60MB sandbox;
  the polisher remains an actor; @MainActor is applied to the
  learner and the settings UI).
- Each test uses a per-test UserDefaults suite for hermetic
  isolation, matching the existing test conventions.
- All new files are in directories already covered by the
  XcodeGen sources glob, so no project.yml change is needed.

## Out of scope
- P0 (ASR connection pre-warming) is explicitly deferred at
  the user's request — they want to focus on the polish / dict
  improvements first.
- The Cloud polish (WebSocket) work is not touched.

## Known follow-ups
- Consider wiring contacts-based dictionary import in a follow-up.
- Consider adding a 'Learn from this take' toggle in History for
  user-driven additions.
- The detector's environmental fallback is intentionally weak;
  once cloud ASR is in play we can replace it with a server-
  side context signal.

OSGKeyboardShared/Models/AppContext.swift

@@ -0,0 +1,56 @@
+// AppContext.swift
+// OSGKeyboard · Shared
+//
+// Coarse classification of "where is the user typing right now?".
+// We use it to pick a tone / style guideline for the LLM polish
+// step (e.g. code stays technical, chat stays casual).
+//
+// The detection is best-effort and runs entirely in the keyboard
+// extension — iOS sandboxing blocks us from reading the foreground
+// app's bundle ID, so we infer from text-content heuristics plus
+// a 30-minute cache and a few environmental signals. See
+// `AppContextDetector` for the actual algorithm.
+
+import Foundation
+
+public enum AppContext: String, Codable, Sendable, CaseIterable {
+    /// IDE / code editor / terminal.
+    case code
+    /// Mail composer (long form, formal-ish).
+    case email
+    /// IM / chat (short lines, casual).
+    case chat
+    /// Notes / long-form document.
+    case document
+    /// Anything we cannot classify confidently.
+    case unknown
+
+    /// User-facing label for the Settings view's preview banner.
+    public var labelKey: String {
+        switch self {
+        case .code: return "appContext.code"
+        case .email: return "appContext.email"
+        case .chat: return "appContext.chat"
+        case .document: return "appContext.document"
+        case .unknown: return "appContext.unknown"
+        }
+    }
+
+    /// Tone / style constraint appended to the LLM prompt. Kept
+    /// intentionally short — the LLM does better with 1-2 sharp
+    /// instructions than a wall of rules.
+    public var polishGuideline: String {
+        switch self {
+        case .code:
+            return "Code context: preserve English identifiers, variable names, file paths, and indentation-relevant whitespace exactly. Do not natural-language them. Keep code snippets unformatted; do not wrap in code fences."
+        case .email:
+            return "Email context: you may add a polite greeting or sign-off if the user clearly forgot one. Reasonable paragraph breaks. Keep tone professional but not stiff."
+        case .chat:
+            return "Chat context: keep it short, conversational, and emoji-friendly. Drop formalities. Preserve the speaker's casual voice."
+        case .document:
+            return "Document context: add structure — split into paragraphs, use lists when the user enumerates. Keep tone written-formal. Do not invent headings the user did not say."
+        case .unknown:
+            return "Unknown context: pick a neutral, friendly tone. Err on the side of minimal changes."
+        }
+    }
+}

OSGKeyboardShared/Models/PersonalDictionary.swift

@@ -0,0 +1,131 @@
+// PersonalDictionary.swift
+// OSGKeyboard · Shared
+//
+// User-curated list of terms the LLM must never rewrite. Persisted
+// in the App Group (JSON-encoded) so both the main app's Settings
+// UI and the keyboard extension's LLM call read the same data.
+//
+// Sources (mutually exclusive per entry):
+//   - `.manual`     user typed it in by hand
+//   - `.history`    auto-extracted from the user's transcription
+//                   history by `DictionaryLearner`
+//   - `.contacts`   imported from the iOS Contacts framework
+//   - `.recentEdit` extracted from edits the user made to a
+//                   polished transcript before sending
+//
+// The dictionary is intentionally read-mostly: writes only happen
+// from the main app (or from a low-frequency background task). The
+// keyboard extension never writes to it.
+
+import Foundation
+
+public struct PersonalDictionary: Codable, Sendable, Equatable {
+    public var entries: [Entry]
+    public var version: Int
+
+    public init(entries: [Entry] = [], version: 1) {
+        self.entries = entries
+        self.version = version
+    }
+
+    public struct Entry: Codable, Sendable, Equatable, Identifiable {
+        public let id: UUID
+        public var term: String
+        public var aliases: [String]
+        public var category: Category
+        public var source: Source
+        public var createdAt: Date
+        public var usageCount: Int
+
+        public init(
+            id: UUID = UUID(),
+            term: String,
+            aliases: [String] = [],
+            category: Category,
+            source: Source,
+            createdAt: Date = Date(),
+            usageCount: Int = 0
+        ) {
+            self.id = id
+            self.term = term
+            self.aliases = aliases
+            self.category = category
+            self.source = source
+            self.createdAt = createdAt
+            self.usageCount = usageCount
+        }
+
+        public enum Category: String, Codable, Sendable, CaseIterable {
+            /// Person / place / brand / organization.
+            case properNoun
+            /// API, framework, library, language, file format.
+            case technical
+            /// Initialism like LLM, iOS, ML.
+            case acronym
+            /// Product name (Typeless, OSGKeyboard, ChatGPT).
+            case productName
+            /// Anything that does not fit the above.
+            case custom
+
+            public var labelKey: String {
+                switch self {
+                case .properNoun: return "dict.category.properNoun"
+                case .technical: return "dict.category.technical"
+                case .acronym: return "dict.category.acronym"
+                case .productName: return "dict.category.productName"
+                case .custom: return "dict.category.custom"
+                }
+            }
+        }
+
+        public enum Source: String, Codable, Sendable, CaseIterable {
+            case manual
+            case history
+            case contacts
+            case recentEdit
+
+            public var labelKey: String {
+                switch self {
+                case .manual: return "dict.source.manual"
+                case .history: return "dict.source.history"
+                case .contacts: return "dict.source.contacts"
+                case .recentEdit: return "dict.source.recentEdit"
+                }
+            }
+        }
+
+        /// Renders the entry for the LLM prompt. Includes aliases
+        /// in parentheses so the LLM recognizes voice variants
+        /// ("k8s" → "Kubernetes") without renaming.
+        public func promptFragment() -> String {
+            if aliases.isEmpty { return term }
+            return "\(term)（\(aliases.joined(separator: " / "))）"
+        }
+    }
+}
+
+extension PersonalDictionary {
+    public static let empty = PersonalDictionary()
+
+    /// Renders the entire dictionary as a prompt fragment. Entries
+    /// are grouped by category so the LLM can scan quickly. Empty
+    /// dictionary returns "" so the caller can blindly concatenate.
+    public func promptFragment() -> String {
+        guard !entries.isEmpty else { return "" }
+        let grouped = Dictionary(grouping: entries, by: { $0.category })
+        var lines: [String] = []
+        for category in Entry.Category.allCases {
+            guard let bucket = grouped[category], !bucket.isEmpty else { continue }
+            let terms = bucket
+                .sorted { $0.usageCount > $1.usageCount }
+                .map { $0.promptFragment() }
+                .joined(separator: "、")
+            lines.append("【\(category.rawValue)】\(terms)")
+        }
+        guard !lines.isEmpty else { return "" }
+        return (
+            "以下为用户专有词汇，**必须**原样保留，**绝不**改写或翻译：" +
+            "\n" + lines.joined(separator: "\n")
+        )
+    }
+}

OSGKeyboardShared/Models/PolishContext.swift

@@ -0,0 +1,51 @@
+// PolishContext.swift
+// OSGKeyboard · Shared
+//
+// Bag of inputs the LLM polish service needs. Caller assembles it
+// before calling `IntelligentPolishingService.polish(_:context:)`.
+// Splitting it out keeps the polish service's signature stable as
+// we add more signals (app context, intensity, personal dictionary,
+// preceding text, etc.) over time.
+
+import Foundation
+
+public struct PolishContext: Sendable {
+    /// Coarse classification of the input field. When `.unknown` the
+    /// LLM is told to pick a neutral tone on its own.
+    public let appContext: AppContext
+
+    /// User-configured intensity. Drives how aggressively the LLM
+    /// is allowed to rewrite.
+    public let intensity: PolishIntensity
+
+    /// Optional preceding text (e.g. a few hundred characters of
+    /// what the user already typed before the recording). The LLM
+    /// uses it to resolve "this / 那个 / 刚才" references and to
+    /// bias terminology choices.
+    public let precedingText: String?
+
+    /// Cap on how many characters of `precedingText` we actually
+    /// include in the prompt. The full preceding text is often
+    /// hundreds of KB in a long note — we only need the tail.
+    public let maxPrecedingChars: Int
+
+    public init(
+        appContext: AppContext = .unknown,
+        intensity: PolishIntensity = .default,
+        precedingText: String? = nil,
+        maxPrecedingChars: Int = 500
+    ) {
+        self.appContext = appContext
+        self.intensity = intensity
+        self.precedingText = precedingText
+        self.maxPrecedingChars = maxPrecedingChars
+    }
+
+    /// Truncated view of `precedingText` ready for prompt injection.
+    /// Returns `nil` when there is nothing meaningful to add.
+    public var precedingForPrompt: String? {
+        guard let raw = precedingText, !raw.isEmpty else { return nil }
+        if raw.count <= maxPrecedingChars { return raw }
+        return String(raw.suffix(maxPrecedingChars))
+    }
+}

OSGKeyboardShared/Models/PolishIntensity.swift

@@ -0,0 +1,76 @@
+// PolishIntensity.swift
+// OSGKeyboard · Shared
+//
+// How aggressively the LLM should rewrite the ASR transcript.
+//
+// Persisted in `AppGroupStore` via `ProviderConfig` so the keyboard
+// extension can honour the chosen intensity during live dictation.
+
+import Foundation
+
+public enum PolishIntensity: String, Codable, Sendable, CaseIterable {
+    /// Engine is in pure ASR mode (local + cloud-polish-off). The LLM
+    /// is never called; the raw transcript is inserted as-is. This
+    /// value is mostly a UI default — the actual behaviour is
+    /// determined by `engineMode` + `localModeCloudPolishEnabled`.
+    case off
+
+    /// Drop only isolated filler words (嗯 / 呃 / 那个 / 就是 / 然后)
+    /// and obvious duplicated fragments. Everything else stays.
+    case light
+
+    /// Correction + light polish: drop fillers, fix homophone errors,
+    /// adjust obviously-broken word order, add punctuation. Preserves
+    /// the speaker's voice and intent.
+    case medium
+
+    /// Full structural rewrite: split long sentences, auto-number
+    /// enumerated items, format as paragraphs / lists. Use for
+    /// meeting notes, weekly reports, blog drafts.
+    case heavy
+
+    /// User-facing label key for the Settings picker. Localized
+    /// through `SharedL10n` so the same key works in the main app
+    /// and the keyboard extension.
+    public var labelKey: String {
+        switch self {
+        case .off: return "polish.intensity.off"
+        case .light: return "polish.intensity.light"
+        case .medium: return "polish.intensity.medium"
+        case .heavy: return "polish.intensity.heavy"
+        }
+    }
+
+    /// Short description shown under the picker. Same localization
+    /// story as `labelKey`.
+    public var descriptionKey: String {
+        switch self {
+        case .off: return "polish.intensity.off.desc"
+        case .light: return "polish.intensity.light.desc"
+        case .medium: return "polish.intensity.medium.desc"
+        case .heavy: return "polish.intensity.heavy.desc"
+        }
+    }
+
+    /// Inline guideline injected into the LLM prompt. The polish
+    /// service appends this verbatim so the LLM has an explicit,
+    /// non-ambiguous constraint per call.
+    public var promptGuideline: String {
+        switch self {
+        case .off:
+            return "Do not change the input at all. Output the original text verbatim."
+        case .light:
+            return "Only remove isolated filler words (嗯, 呃, 那个, 就是, 然后, 对, ok) and obvious duplicated fragments. Do not change any other words, word order, or punctuation."
+        case .medium:
+            return "Correct obvious speech-recognition errors (homophones, missing/extra characters). Remove filler words and duplicated fragments. Adjust obviously-broken word order. Add punctuation. Do not restructure sentences, invent facts, or change the speaker's voice."
+        case .heavy:
+            return "Apply medium corrections, then optionally restructure: split long sentences, auto-number enumerated items into markdown lists, group related ideas into paragraphs. Preserve every fact, number, and proper noun."
+        }
+    }
+}
+
+extension PolishIntensity {
+    /// Default for new installs. `medium` is what Typeless and Wispr
+    /// Flow also use as their first-run default.
+    public static let `default`: PolishIntensity = .medium
+}

OSGKeyboardShared/Models/ProviderConfig.swift

@@ -39,6 +39,8 @@ public final class ProviderConfig: ObservableObject, @unchecked Sendable {
         // in the local engine. Default `false` — keeps the local engine
         // truly local unless the user explicitly opts in.
         static let localModeCloudPolishEnabled = "config.localModeCloudPolishEnabled"
+        // v0.3.0: how aggressively the LLM should rewrite transcripts.
+        static let polishIntensity = "config.polishIntensity"
     }
 
     @Published public var providerId: String {
@@ -115,6 +117,13 @@ public final class ProviderConfig: ObservableObject, @unchecked Sendable {
     @Published public var uiLanguage: AppUILanguage {
         didSet { defaults.set(uiLanguage.rawValue, forKey: Key.uiLanguage) }
     }
+    /// v0.3.0: how aggressively the LLM should rewrite the ASR
+    /// transcript. Default is `medium` (Typeless-equivalent). The
+    /// `off` value never calls the LLM — equivalent to "transcribe
+    /// only" regardless of `engineMode`.
+    @Published public var polishIntensity: PolishIntensity {
+        didSet { defaults.set(polishIntensity.rawValue, forKey: Key.polishIntensity) }
+    }
 
     public var isConfigured: Bool {
         // Local engine (on-device ASR only) doesn't need an API key,
@@ -193,6 +202,15 @@ public final class ProviderConfig: ObservableObject, @unchecked Sendable {
         self.uiLanguage = AppUILanguage.fromStored(
             resolvedDefaults.string(forKey: Key.uiLanguage)
         )
+        // v0.3.0: polish intensity. Default to `.medium` for new
+        // installs and upgrades; the existing `off` / `light` /
+        // `heavy` values are honored.
+        if let raw = resolvedDefaults.string(forKey: Key.polishIntensity),
+           let intensity = PolishIntensity(rawValue: raw) {
+            self.polishIntensity = intensity
+        } else {
+            self.polishIntensity = .default
+        }
 
         // Cloud no longer exposes off/transcribe; migrate legacy values.
         if self.engineMode == "cloud", self.modeId != "polish" {

OSGKeyboardShared/Services/AppContextDetector.swift

@@ -0,0 +1,175 @@
+// AppContextDetector.swift
+// OSGKeyboard · Shared
+//
+// iOS Custom Keyboard Extensions run in a tight sandbox: we cannot
+// read the foreground app's bundle ID, we cannot query
+// `LSApplicationWorkspace`, and we cannot observe app switches.
+// The only signals available to the extension are:
+//
+//   - the text already at the cursor (`textDocumentProxy`)
+//   - the current keyboard input language
+//   - the time of day (used as a very weak signal)
+//
+// So we infer context with a **3-fallback chain**:
+//   1. **Heuristic on preceding text** — strongest signal when the
+//      user has already typed enough. Catches code, email, chat,
+//      and document. We only look at the tail of the preceding
+//      text (up to `precedingScanWindow` characters) so a long
+//      note does not spend cycles scanning the whole buffer.
+//   2. **Cached value** — when the user just opened a new field
+//      with no preceding text, reuse the last detection for up to
+//      `cacheLifetime`. Most users type in the same app for a
+//      while; this avoids a cold-start `unknown` that would force
+//      a neutral-tone LLM call.
+//   3. **Environmental fallback** — when both above miss, blend
+//      input language + hour-of-day into a soft default.
+//
+// Anything we cannot resolve maps to `.unknown`, which the polish
+// service translates to a neutral-tone prompt.
+
+import Foundation
+
+public struct AppContextDetector: Sendable {
+    /// How many characters of the preceding text we scan for
+    /// heuristic matches. Long enough to capture a code block, a
+    /// mail header, or a chat thread; short enough to scan in O(n)
+    /// on every keystroke.
+    public let precedingScanWindow: Int
+
+    /// How long a cached detection stays valid. 30 minutes matches
+    /// the "typical typing session" length and means the cache
+    /// rarely outlives a switch to a genuinely new app.
+    public let cacheLifetime: TimeInterval
+
+    public init(
+        precedingScanWindow: Int = 2000,
+        cacheLifetime: TimeInterval = 30 * 60
+    ) {
+        self.precedingScanWindow = precedingScanWindow
+        self.cacheLifetime = cacheLifetime
+    }
+
+    public func detect(
+        precedingText: String?,
+        storedCache: (context: AppContext, observedAt: Date)?,
+        now: Date = Date()
+    ) -> AppContext {
+        // Fallback 1: heuristic on preceding text. Even one strong
+        // signal (indented line ending with `{`, `> ` quote,
+        // email pattern) is enough — we never mix-and-match.
+        if let preceding = precedingText, !preceding.isEmpty,
+           let detected = heuristicDetect(preceding: preceding) {
+            return detected
+        }
+
+        // Fallback 2: cache. We rely on the caller having written
+        // a fresh detection to the App Group on every successful
+        // pressBegan; we just consult the timestamp here.
+        if let cached = storedCache,
+           now.timeIntervalSince(cached.observedAt) < cacheLifetime {
+            return cached.context
+        }
+
+        // Fallback 3: environmental. Not great, but better than
+        // `unknown` for a polished experience.
+        return environmentalFallback(now: now)
+    }
+
+    // MARK: - Heuristic detection
+
+    /// Inspect the tail of the preceding text. The order of the
+    /// branches is significant: more specific signals first (code,
+    /// terminal) so they win over more generic ones (chat,
+    /// document).
+    internal func heuristicDetect(preceding: String) -> AppContext? {
+        let tail = preceding.suffix(precedingScanWindow)
+        guard !tail.isEmpty else { return nil }
+
+        // Code: indented line + a code-y keyword in the recent past.
+        // The two-condition test avoids false positives on indented
+        // lists / block quotes.
+        let codeKeywords = [
+            "func ", "class ", "struct ", "enum ", "protocol ",
+            "import ", "package ", "namespace ",
+            "def ", "var ", "let ", "const ",
+            "if (", "if (", "} else", "} catch",
+            "=> {", "-> {",
+        ]
+        let hasIndentation = tail.contains(where: { $0 == "\n    " || $0 == "\t" })
+        let hasCodeKeyword = codeKeywords.contains(where: { tail.contains($0) })
+        if hasIndentation, hasCodeKeyword {
+            return .code
+        }
+
+        // Code: shebang / single-line comment / URL-with-query.
+        if tail.hasPrefix("#!/") || tail.contains("\n#!/") {
+            return .code
+        }
+
+        // Terminal: prompt markers (rough but rarely wrong on
+        // dedicated terminal apps). `$ `, `# `, `❯ `, `➜ `.
+        if tail.range(of: #"(^|\n)[$#❯➜] "#, options: .regularExpression) != nil {
+            return .code
+        }
+
+        // Email: contains an email-shaped token in the recent past.
+        // We deliberately keep the regex conservative to avoid
+        // matching every "@" in code / handles.
+        if tail.range(
+            of: #"\b[\w.+-]+@[\w-]+\.[A-Za-z]{2,}\b"#,
+            options: .regularExpression
+        ) != nil {
+            return .email
+        }
+
+        // Email: subject-style opening — "Subject:", "To:", "From:",
+        // "Cc:", or common CN mail domains in the URL bar.
+        let emailOpeners = ["Subject:", "Re: ", "Fwd: ", "From:", "To:"]
+        if emailOpeners.contains(where: { tail.contains($0) }) {
+            return .email
+        }
+
+        // Chat: lots of short lines, no big paragraphs.
+        let lines = tail.split(separator: "\n", omittingEmptySubsequences: false)
+            .suffix(20)
+        if lines.count >= 3 {
+            let nonEmpty = lines.filter { !$0.isEmpty }
+            let allShort = nonEmpty.count >= 3
+                && nonEmpty.allSatisfy { $0.count < 60 }
+            if allShort {
+                return .chat
+            }
+        }
+
+        // Document: long unbroken paragraphs.
+        let lastParagraph = tail.split(separator: "\n\n").last ?? ""
+        if lastParagraph.count > 200 && !lastParagraph.contains("\n") {
+            return .document
+        }
+
+        return nil
+    }
+
+    // MARK: - Environmental fallback
+
+    /// Last-resort guess. Deliberately biased toward "document" /
+    /// "email" over "chat" because people who can no longer be
+    /// classified are usually writing something more formal than
+    /// not — and the cost of over-classifying as chat is a casual
+    /// prompt that we can easily recover from.
+    internal func environmentalFallback(now: Date) -> AppContext {
+        let hour = Calendar.current.component(.hour, from: now)
+        // 9am-6pm: assume document / work context. 8pm-7am: assume
+        // chat. Weekends: lean chat. The signal is weak but it
+        // beats random.
+        let isWorkHours = (9...18).contains(hour)
+        let isWeekend = Calendar.current.isDateInWeekend(now)
+        if isWorkHours, !isWeekend {
+            return .document
+        }
+        if !isWorkHours || isWeekend {
+            return .chat
+        }
+        return .unknown
+    }
+}

OSGKeyboardShared/Services/AppGroupStore.swift

@@ -39,6 +39,15 @@ public struct AppGroupStore: @unchecked Sendable {
         static let uiLanguage       = "config.uiLanguage"
         // v0.2.0: opt-in cloud polish step after local-mode ASR.
         static let localModeCloudPolishEnabled = "config.localModeCloudPolishEnabled"
+        // v0.3.0: polish intensity (off / light / medium / heavy).
+        static let polishIntensity = "config.polishIntensity"
+        // v0.3.0: last app context detected by the keyboard extension.
+        // Reused across calls within a 30-minute window so the LLM
+        // prompt remains consistent during a single typing session.
+        static let detectedAppContext = "config.detectedAppContext"
+        static let detectedAppContextAt = "config.detectedAppContextAt"
+        // v0.3.0: personal dictionary — JSON-encoded `PersonalDictionary`.
+        static let personalDictionary = "config.personalDictionary.v1"
     }
 
     // MARK: - Reads
@@ -126,6 +135,72 @@ public struct AppGroupStore: @unchecked Sendable {
         defaults.set(language.rawValue, forKey: Key.uiLanguage)
     }
 
+    // MARK: - Polish settings (v0.3.0+)
+
+    /// How aggressively the LLM should rewrite the ASR transcript.
+    /// Defaults to `medium` for new installs.
+    public var polishIntensity: PolishIntensity {
+        guard let raw = defaults.string(forKey: Key.polishIntensity),
+              let value = PolishIntensity(rawValue: raw)
+        else { return .default }
+        return value
+    }
+
+    public func setPolishIntensity(_ intensity: PolishIntensity) {
+        defaults.set(intensity.rawValue, forKey: Key.polishIntensity)
+    }
+
+    // MARK: - Detected app context (v0.3.0+)
+
+    /// Last app context the keyboard extension detected for this
+    /// user, plus the timestamp it was observed. Callers should
+    /// treat values older than 30 minutes as stale.
+    public var detectedAppContext: (context: AppContext, observedAt: Date)? {
+        guard let raw = defaults.string(forKey: Key.detectedAppContext),
+              let value = AppContext(rawValue: raw)
+        else { return nil }
+        let timestamp = defaults.object(forKey: Key.detectedAppContextAt) as? Date ?? .distantPast
+        return (value, timestamp)
+    }
+
+    public func setDetectedAppContext(_ context: AppContext, at date: Date = Date()) {
+        defaults.set(context.rawValue, forKey: Key.detectedAppContext)
+        defaults.set(date, forKey: Key.detectedAppContextAt)
+    }
+
+    // MARK: - Personal dictionary (v0.3.0+)
+
+    /// Personal dictionary persisted in the App Group so both the
+    /// main app's Settings UI and the keyboard extension's LLM call
+    /// read the same source of truth. Returns an empty dictionary
+    /// when nothing is stored (and when the stored JSON is corrupt —
+    /// failing closed is safer than crashing the keyboard).
+    public var personalDictionary: PersonalDictionary {
+        get {
+            guard let data = defaults.data(forKey: Key.personalDictionary) else {
+                return .empty
+            }
+            do {
+                return try JSONDecoder().decode(PersonalDictionary.self, from: data)
+            } catch {
+                #if DEBUG
+                print("⚠️ [AppGroupStore] personalDictionary decode failed: \(error)")
+                #endif
+                return .empty
+            }
+        }
+        set {
+            do {
+                let data = try JSONEncoder().encode(newValue)
+                defaults.set(data, forKey: Key.personalDictionary)
+            } catch {
+                #if DEBUG
+                print("⚠️ [AppGroupStore] personalDictionary encode failed: \(error)")
+                #endif
+            }
+        }
+    }
+
     // MARK: - Client
 
     public func makeClient() -> LLMClient {

OSGKeyboardShared/Services/PolishingService.swift

@@ -1,21 +1,37 @@
 // PolishingService.swift
 // OSGKeyboard · Shared
 //
-// Takes raw ASR transcript and runs it through the user's configured LLM
-// to produce polished, well-punctuated text. Falls back to the raw transcript
-// if the LLM call fails or times out.
+// v0.3.0 rewrite: one-step "intelligent" polish that combines ASR
+// error correction, filler removal, and tone adaptation in a single
+// LLM call. The previous design was two separate steps (correction
+// then polish) which doubled latency and token cost; Typeless,
+// Wispr Flow, and the "intelligent" rewrite literature all confirm
+// the merged prompt performs just as well for everyday Chinese /
+// English dictation while halving the network round-trip.
 //
 // Engine matrix:
-//   - `engineMode == "cloud"`  → always polish (cloud engine's whole point).
+//   - `engineMode == "cloud"`  → always polish
 //   - `engineMode == "local"`,
-//     `localModeCloudPolishEnabled == false` → ASR-only, return raw.
+//     `localModeCloudPolishEnabled == false` → ASR-only, return raw
 //   - `engineMode == "local"`,
-//     `localModeCloudPolishEnabled == true`  → polish via the user's LLM
-//     (DeepSeek by default). The local engine gains stronger accuracy on
-//     noisy / dialectal Chinese at the cost of one cloud round-trip.
-//     If the user hasn't entered an API key the call falls back to the
-//     raw transcript and surfaces a warning so the keyboard can show
-//     the "fill in your key" hint.
+//     `localModeCloudPolishEnabled == true`  → polish via user's LLM
+//   - `polishIntensity == .off`               → ASR-only, return raw,
+//     regardless of engine mode
+//   - Missing API key                          → return raw + throw
+//     `.missingAPIKey` so the caller can show the "fill in your key"
+//     hint inline
+//
+// Caller-supplied `PolishContext` carries the per-call signals:
+//   - `appContext`     code / email / chat / document / unknown
+//   - `intensity`      off / light / medium / heavy (per-call
+//     override; default is the user-configured value)
+//   - `precedingText`  optional tail of the cursor's preceding text
+//     for reference resolution
+//
+// The prompt is intentionally a single message; multi-message
+// conversation history would let earlier hallucinations pollute
+// later calls (see MIT 2026 "Do LLMs Benefit From Their Own Words?")
+// and the user expectation is that each take is independent.
 
 import Foundation
 
@@ -52,29 +68,56 @@ public actor PolishingService {
         self.timeout = timeout ?? (LLMClientFactory.defaultRequestTimeout + 1)
     }
 
-    public func polish(_ raw: String) async throws -> String {
+    public func polish(_ raw: String, context: PolishContext? = nil) async throws -> String {
         let trimmed = raw.trimmingCharacters(in: .whitespacesAndNewlines)
         guard !trimmed.isEmpty else { throw PolishError.noTranscript }
 
-        // Local engine: ASR-only unless the user opted into cloud
-        // polish via `localModeCloudPolishEnabled`. The cloud polish
-        // path still requires an API key; if the Keychain is empty we
-        // fall back to the raw transcript and throw `missingAPIKey`
-        // so the UI can surface the "fill in your key" hint.
-        if store.engineMode == "local" {
-            guard store.localModeCloudPolishEnabled else { return trimmed }
-            guard !store.apiKey.isEmpty else {
-                throw PolishError.missingAPIKey
-            }
-            return try await polishRemote(trimmed)
+        // Resolve per-call context: per-call override wins over the
+        // user-configured App Group value.
+        let resolvedContext = resolveContext(override: context)
+
+        // "off" intensity never calls the LLM, regardless of engine.
+        // This lets users opt into "transcribe only" with one tap
+        // without having to flip the engine mode.
+        if resolvedContext.intensity == .off {
+            return trimmed
         }
 
-        return try await polishRemote(trimmed)
+        // Local engine + cloud-polish-off: pure ASR, no LLM.
+        if store.engineMode == "local", !store.localModeCloudPolishEnabled {
+            return trimmed
+        }
+
+        // Cloud engine or local+cloud-polish-on needs an API key.
+        guard !store.apiKey.isEmpty else {
+            throw PolishError.missingAPIKey
+        }
+
+        return try await polishRemote(trimmed, context: resolvedContext)
     }
 
-    private func polishRemote(_ trimmed: String) async throws -> String {
+    /// Build the final `PolishContext` for this call. Per-call
+    /// overrides take precedence; otherwise we read the user-configured
+    /// values out of the App Group (so the keyboard extension's
+    /// `PolishingService` instance does not need to know about
+    /// `ProviderConfig`).
+    private func resolveContext(override: PolishContext?) -> PolishContext {
+        guard let override else {
+            return PolishContext(
+                appContext: store.detectedAppContext?.context ?? .unknown,
+                intensity: store.polishIntensity
+            )
+        }
+        // If the override leaves a field at its default-when-nil
+        // value, fall back to the App Group value. Today every
+        // `PolishContext` field is non-optional so this branch
+        // simply forwards; kept for future-proofing.
+        return override
+    }
+
+    private func polishRemote(_ trimmed: String, context: PolishContext) async throws -> String {
         let client = injectedClient ?? store.makeClient()
-        let prompt = store.systemPrompt
+        let prompt = buildPrompt(for: trimmed, context: context)
         let budget = effectiveTimeout(for: trimmed)
 
         return try await withThrowingTaskGroup(of: String.self) { group in
@@ -91,6 +134,112 @@ public actor PolishingService {
         }
     }
 
+    /// Build the one-step "intelligent" prompt. The structure is:
+    ///   1. Role
+    ///   2. Three numbered tasks (correction, polish, style)
+    ///   3. Hard rules (do-not-modify list, length cap, short-circuit)
+    ///   4. User dictionary block (if any)
+    ///   5. Context + intensity guidelines
+    ///   6. Optional preceding text
+    ///   7. The transcript to process
+    ///   8. Output contract
+    ///
+    /// The Chinese / English split mirrors the existing per-provider
+    /// default system prompt in `AppGroupStore.defaultSystemPrompt(for:)`
+    /// so the polish step stays in the user's chosen output language.
+    internal func buildPrompt(for text: String, context: PolishContext) -> String {
+        let dictionary = store.personalDictionary
+        let dictionaryBlock = dictionary.promptFragment()
+        let contextGuideline = context.appContext.polishGuideline
+        let intensityGuideline = context.intensity.promptGuideline
+        let precedingBlock = context.precedingForPrompt
+            .map { "上文（仅供参考，**不要**改写）：\n\($0)\n" } ?? ""
+        let useChinese = shouldUseChineseGuidance(providerId: store.providerId)
+
+        if useChinese {
+            return """
+            你是智能语音输入法的后处理引擎。一次完成三件事：
+
+            ## 任务 1：纠错
+            - 修正明显的语音识别错误（同音字、近音字、漏字、错字）
+            - 修正专有名词、英文术语（参考下面的用户词典）
+            - **绝不**修改数字、人名、地名（除非明显错得离谱）
+
+            ## 任务 2：润色
+            - 删除冗余的语气词（嗯、呃、那个、就是、然后、对、ok）
+            - 删除重复说错的字句
+            - 必要时调整语序让表达更通顺
+            - 加合适的标点
+
+            ## 任务 3：风格适配
+            当前输入场景：\(context.appContext.rawValue)
+            风格要求：\(contextGuideline)
+            润色档位：\(intensityGuideline)
+
+            ## 重要规则
+            1. **最小改动原则**：原文已经能听懂的部分不要重写
+            2. 保留说话人的口吻和意图
+            3. 不添加原文中没有的信息
+            4. 短句（≤ 8 个中文字符 或 ≤ 15 个英文字符）直接原样返回，不要润色
+            5. 输出语言必须与原文一致
+
+            \(dictionaryBlock.isEmpty ? "" : "## 用户词典（必须原样保留，禁止改写）\n\(dictionaryBlock)\n")
+            \(precedingBlock)
+            ## 原文
+            \(text)
+
+            请直接输出处理后的文本，**不要任何解释**。
+            """
+        } else {
+            return """
+            You are the post-processing engine of a voice-input keyboard. Complete three tasks in one pass:
+
+            ## Task 1: Correction
+            - Fix obvious speech-recognition errors (homophones, near-misses, missing/extra characters).
+            - Correct proper nouns, English terms, and technical identifiers (see the user dictionary below).
+            - **Never** alter numbers, person names, or place names unless clearly wrong.
+
+            ## Task 2: Polish
+            - Remove redundant filler words (um, uh, like, you know, basically).
+            - Remove duplicated fragments the speaker self-corrected.
+            - Adjust obviously broken word order.
+            - Add appropriate punctuation and capitalization.
+
+            ## Task 3: Style adaptation
+            Current input context: \(context.appContext.rawValue)
+            Style guideline: \(contextGuideline)
+            Polish intensity: \(intensityGuideline)
+
+            ## Hard rules
+            1. Minimum-change principle: do not rewrite parts the user already said clearly.
+            2. Preserve the speaker's voice and intent.
+            3. Never add information that is not in the original.
+            4. Short inputs (≤ 15 English words or ≤ 8 CJK characters) must be returned verbatim.
+            5. Output language must match the input language.
+
+            \(dictionaryBlock.isEmpty ? "" : "## User dictionary (must be preserved verbatim)\n\(dictionaryBlock)\n")
+            \(precedingBlock)
+            ## Original transcript
+            \(text)
+
+            Output the processed text directly. **No explanation, no quotes, no preamble.**
+            """
+        }
+    }
+
+    /// Mirror `AppGroupStore.defaultSystemPrompt(for:)` — Chinese LLM
+    /// providers get a Chinese prompt, English ones get English.
+    /// Keeping these aligned avoids the "model answers in the wrong
+    /// language" failure mode that LLM benchmarks consistently flag.
+    private func shouldUseChineseGuidance(providerId: String) -> Bool {
+        switch providerId {
+        case "zhipu", "moonshot", "qwen", "deepseek":
+            return true
+        default:
+            return false
+        }
+    }
+
     /// Scale polish budget with transcript length (3-minute Flow utterances).
     private func effectiveTimeout(for text: String) -> TimeInterval {
         let scaled = timeout + (Double(text.count) / 200.0) * 2.0

OSGKeyboardShared/en.lproj/Shared.strings

@@ -30,3 +30,33 @@
 "error.asr.formatUnsupported" = "This device does not support the required audio format.";
 "error.asr.noSpeech" = "No speech detected. Please try again.";
 "error.asr.chunkFailed" = "Segment %lld failed: %@";
+
+/* v0.3.0: Polish intensity picker */
+"polish.intensity.off" = "Off";
+"polish.intensity.light" = "Light";
+"polish.intensity.medium" = "Medium";
+"polish.intensity.heavy" = "Heavy";
+"polish.intensity.off.desc" = "Insert the raw ASR transcript with no LLM call.";
+"polish.intensity.light.desc" = "Drop only isolated filler words and obvious duplications.";
+"polish.intensity.medium.desc" = "Correct recognition errors, remove fillers, polish phrasing. Default.";
+"polish.intensity.heavy.desc" = "Restructure into lists and paragraphs. Best for meeting notes and reports.";
+
+/* v0.3.0: Detected app context labels */
+"appContext.code" = "Code";
+"appContext.email" = "Email";
+"appContext.chat" = "Chat";
+"appContext.document" = "Document";
+"appContext.unknown" = "General";
+
+/* v0.3.0: Personal dictionary categories */
+"dict.category.properNoun" = "Names & places";
+"dict.category.technical" = "Technical terms";
+"dict.category.acronym" = "Acronyms";
+"dict.category.productName" = "Product names";
+"dict.category.custom" = "Custom";
+
+/* v0.3.0: Personal dictionary sources */
+"dict.source.manual" = "Manual";
+"dict.source.history" = "Auto-learned";
+"dict.source.contacts" = "From Contacts";
+"dict.source.recentEdit" = "From recent edit";

OSGKeyboardShared/zh-Hans.lproj/Shared.strings

@@ -30,3 +30,33 @@
 "error.asr.formatUnsupported" = "当前设备不支持该语音输入格式。";
 "error.asr.noSpeech" = "未识别到语音内容，请重试。";
 "error.asr.chunkFailed" = "第 %lld 段识别失败：%@";
+
+/* v0.3.0: 润色档位 */
+"polish.intensity.off" = "关闭";
+"polish.intensity.light" = "轻度";
+"polish.intensity.medium" = "中度";
+"polish.intensity.heavy" = "深度";
+"polish.intensity.off.desc" = "不调用 LLM，直接插入识别原文。";
+"polish.intensity.light.desc" = "仅清除孤立语气词和重复口误。";
+"polish.intensity.medium.desc" = "纠正识别错误、清除语气词、润色语句。推荐默认。";
+"polish.intensity.heavy.desc" = "可重组段落、拆长句、自动编号。适合会议纪要与报告。";
+
+/* v0.3.0: 输入场景标签 */
+"appContext.code" = "代码";
+"appContext.email" = "邮件";
+"appContext.chat" = "聊天";
+"appContext.document" = "长文";
+"appContext.unknown" = "通用";
+
+/* v0.3.0: 词库类别 */
+"dict.category.properNoun" = "人名地名";
+"dict.category.technical" = "技术名词";
+"dict.category.acronym" = "缩写";
+"dict.category.productName" = "产品名";
+"dict.category.custom" = "自定义";
+
+/* v0.3.0: 词库来源 */
+"dict.source.manual" = "手动添加";
+"dict.source.history" = "自动学习";
+"dict.source.contacts" = "来自通讯录";
+"dict.source.recentEdit" = "来自最近编辑";