feat: add 6 Boost VC AI capability domains

Cargo.toml

@@ -114,6 +114,7 @@ dotenvy = "0.15"
 console = "0.16"
 regex = "1.10"
 walkdir = "2"
+sqlparser = "0.62"
 glob = "0.3"
 unicode-segmentation = "1"
 unicode-width = "0.2"
@@ -126,7 +127,7 @@ sysinfo = { version = "0.33", default-features = false, features = ["system"] }
 keyring = { version = "3", features = ["apple-native", "windows-native", "linux-native"] }
 clap = { version = "4.5", features = ["derive"] }
 clap_complete = "4.5"
-lettre = { version = "0.11.22", default-features = false, features = ["builder", "smtp-transport", "rustls-tls"] }
+lettre = { version = "0.11.22", default-features = false, features = ["builder", "smtp-transport", "tokio1-rustls-tls"] }
 mail-parser = "0.11.2"
 async-imap = { version = "0.11", features = ["runtime-tokio"], default-features = false }
 axum = { version = "0.8", default-features = false, features = ["http1", "json", "tokio", "query", "ws", "macros"] }
@@ -154,6 +155,9 @@ fs2 = "0.4"
 starship-battery = "0.10"
 ethers-core = { version = "2.0.14", default-features = false }
 ethers-signers = { version = "2.0.14", default-features = false }
+whatlang = "0.18"
+nnnoiseless = "0.5"
+strsim = "0.11"
 # Multi-chain wallet signing.
 # - bitcoin: P2WPKH PSBT build/sign/broadcast (includes secp256k1).
 # - ed25519-dalek: Solana transaction signing.

docs/boost-vc-capability-plan.md

@@ -0,0 +1,77 @@
+# Boost VC AI Capability Plan
+
+## Commercial Inspirations
+
+| Capability | Inspiration | What we replicate |
+|-----------|-------------|-------------------|
+| Voice Foundation | Siri, Google Assistant | Local STT (Whisper) + TTS (Piper) desktop assistant |
+| Live Captions | Otter.ai, Microsoft Teams captions | Real-time transcription with saved transcripts |
+| Voice Actions | Alexa Skills, Siri Shortcuts | Utterance → controller-backed action routing |
+| Operator Inbox | Front, Superhuman | Triage, draft replies, follow-up scheduling |
+| Chat-with-Data | Julius AI, ChatGPT Code Interpreter | NL queries over local/connected datasets |
+| Guided Recommendations | Typeform, Intercom Product Tours | Quiz-style intake flows with branching logic |
+
+## Features Replicated (v1)
+
+### Voice Foundation (#1831)
+- Session lifecycle (start/stop/status)
+- PCM buffering with VAD (voice activity detection)
+- STT via whisper-rs (local, open-source)
+- TTS via Piper (local, open-source)
+- LLM turn orchestration (STT → LLM → TTS)
+- Conversation history context
+
+### Live Captions (#1832)
+- Transcript lifecycle (start/pause/resume/complete)
+- Real-time segment appending with timestamps
+- Extractive summarization on completed transcripts
+- Source-agnostic (microphone or desktop audio)
+
+### Voice Actions (#1833)
+- Action registration with trigger phrases
+- Fuzzy intent recognition (word overlap scoring)
+- Safety levels (safe/confirmation_required/destructive)
+- Confirmation flow for non-safe actions
+- Execution tracking with status
+
+### Operator Inbox (#1834)
+- Priority scoring (urgent/high/medium/low)
+- Multi-tone draft generation (professional/casual/formal)
+- Follow-up scheduling
+- Archive workflow
+
+### Chat-with-Data (#1835)
+- Dataset registration (CSV, database, API sources)
+- Natural language query routing
+- Proactive insight generation (anomaly detection)
+- Dataset listing and metadata
+
+### Guided Recommendations (#1836)
+- Flow definition with branching steps
+- Answer validation (type checking, choice validation)
+- State machine (active → completed)
+- Recommendation generation based on answers
+- Builtin onboarding setup flow
+
+## Explicit Non-Goals (v1)
+
+- **No real-time streaming STT** — batch transcription per VAD segment only
+- **No speaker diarization** — single-speaker assumption for v1
+- **No actual email/Slack integration** — operator inbox is schema-only, no transport
+- **No real SQL execution** — chat-with-data generates mock query results
+- **No ML-based intent recognition** — word overlap heuristic, not a trained model
+- **No persistent storage** — all state is in-memory (process-lifetime)
+- **No frontend components** — backend domain modules only, frontend wiring is follow-up
+- **No multi-language TTS** — English-only for Piper in v1
+
+## Architecture
+
+All capabilities follow the same pattern:
+- Rust domain module under `src/openhuman/<domain>/`
+- `types.rs` — domain types with serde
+- `engine.rs` — business logic + state machine
+- `rpc.rs` — JSON-RPC handlers
+- `schemas.rs` — controller registry schemas (Def pattern)
+- Wired into `core/all.rs` (controller registry + namespace description)
+- Catalog entry in `about_app/catalog.rs`
+- Structured tracing (`debug!`/`info!`/`warn!`) at all state transitions

src/core/all.rs

@@ -251,6 +251,21 @@ fn build_registered_controllers() -> Vec<RegisteredController> {
     controllers.extend(
         crate::openhuman::desktop_companion::all_desktop_companion_registered_controllers(),
     );
+    // Standalone voice assistant: local STT/TTS conversational loop.
+    controllers
+        .extend(crate::openhuman::voice_assistant::all_voice_assistant_registered_controllers());
+    // Guided recommendation flows: quiz-style intake and recommendation engine.
+    controllers.extend(crate::openhuman::guided_flows::all_guided_flows_registered_controllers());
+    // Live captions: real-time transcription, transcript storage, and summarization.
+    controllers.extend(crate::openhuman::live_captions::all_live_captions_registered_controllers());
+    // Voice actions: intent recognition and controller-backed action execution.
+    controllers.extend(crate::openhuman::voice_actions::all_voice_actions_registered_controllers());
+    // Operator inbox: message triage, draft generation, and follow-up scheduling.
+    controllers
+        .extend(crate::openhuman::operator_inbox::all_operator_inbox_registered_controllers());
+    // Chat-with-data: NL querying over datasets and proactive insight generation.
+    controllers
+        .extend(crate::openhuman::chat_with_data::all_chat_with_data_registered_controllers());
     // Structured WhatsApp Web data — agent-facing read-only controllers (list/search).
     // The write-path ingest controller is registered separately in build_internal_only_controllers.
     controllers.extend(crate::openhuman::whatsapp_data::all_whatsapp_data_registered_controllers());
@@ -362,6 +377,18 @@ fn build_declared_controller_schemas() -> Vec<ControllerSchema> {
     schemas.extend(crate::openhuman::desktop_companion::all_desktop_companion_controller_schemas());
     // Structured WhatsApp Web data — local SQLite store, agent-queryable
     schemas.extend(crate::openhuman::whatsapp_data::all_whatsapp_data_controller_schemas());
+    // Standalone voice assistant
+    schemas.extend(crate::openhuman::voice_assistant::all_voice_assistant_controller_schemas());
+    // Guided recommendation flows
+    schemas.extend(crate::openhuman::guided_flows::all_guided_flows_controller_schemas());
+    // Live captions and transcripts
+    schemas.extend(crate::openhuman::live_captions::all_live_captions_controller_schemas());
+    // Voice-driven actions
+    schemas.extend(crate::openhuman::voice_actions::all_voice_actions_controller_schemas());
+    // Operator inbox triage
+    schemas.extend(crate::openhuman::operator_inbox::all_operator_inbox_controller_schemas());
+    // Chat-with-data analytics
+    schemas.extend(crate::openhuman::chat_with_data::all_chat_with_data_controller_schemas());
     // Mobile device pairing and management
     schemas.extend(crate::openhuman::devices::all_devices_controller_schemas());
     schemas
@@ -486,6 +513,24 @@ pub fn namespace_description(namespace: &str) -> Option<&'static str> {
         "companion" => Some(
             "Desktop companion — Clicky-style hotkey-driven interaction loop with STT, LLM, TTS, and visual pointing.",
         ),
+        "voice_assistant" => Some(
+            "Standalone local-first voice assistant — mic → VAD → STT → LLM → TTS → speaker loop with session management.",
+        ),
+        "guided_flows" => Some(
+            "Reusable guided recommendation and intake flows — quiz-style state machine with branching, validation, and recommendation generation.",
+        ),
+        "live_captions" => Some(
+            "Real-time captioning, transcript persistence, and meeting-note summarization from microphone or system audio.",
+        ),
+        "voice_actions" => Some(
+            "Voice-driven desktop actions — maps utterances to controller-backed commands with safety levels and confirmation flows.",
+        ),
+        "operator_inbox" => Some(
+            "Operator inbox assistant — message triage, priority scoring, draft reply generation, and follow-up scheduling.",
+        ),
+        "chat_with_data" => Some(
+            "Chat-with-data analytics — natural-language querying over datasets with proactive insight and anomaly detection.",
+        ),
         _ => None,
     }
 }

src/openhuman/about_app/catalog.rs

@@ -1230,6 +1230,106 @@ const CAPABILITIES: &[Capability] = &[
             destinations: &["Google Meet", "ElevenLabs (STT/TTS via hosted backend)"],
         }),
     },
+    // ── Voice Assistant ─────────────────────────────────────────────────────
+    Capability {
+        id: "voice_assistant.session",
+        name: "Standalone Voice Assistant",
+        domain: "voice_assistant",
+        category: CapabilityCategory::Automation,
+        description: "A local-first conversational voice assistant that uses free/open STT \
+                      (Whisper via whisper.cpp) and TTS (Piper) to provide hands-free desktop \
+                      interaction. Mic audio is processed locally by default; cloud fallback \
+                      available when configured.",
+        how_to: "Start a voice session via RPC: openhuman.voice_assistant_start_session.",
+        status: CapabilityStatus::Beta,
+        privacy: Some(CapabilityPrivacy {
+            leaves_device: true,
+            data_kind: PrivacyDataKind::UserContent,
+            destinations: &["OpenHuman backend", "TinyHumans Neocortex"],
+        }),
+    },
+    // ── Guided Flows ────────────────────────────────────────────────────────
+    Capability {
+        id: "guided_flows.recommendation",
+        name: "Guided Recommendation Flows",
+        domain: "guided_flows",
+        category: CapabilityCategory::Automation,
+        description: "Reusable quiz-style or conversational intake flows that guide users to \
+                      recommendations, decisions, or next actions. Includes a built-in onboarding \
+                      setup guide with branching logic and rule-based recommendation generation.",
+        how_to: "Start a flow via RPC: openhuman.guided_flows_start_flow with flow_id.",
+        status: CapabilityStatus::Beta,
+        privacy: Some(CapabilityPrivacy {
+            leaves_device: true,
+            data_kind: PrivacyDataKind::UserContent,
+            destinations: &["OpenHuman backend", "TinyHumans Neocortex"],
+        }),
+    },
+    // ── Live Captions ───────────────────────────────────────────────────────
+    Capability {
+        id: "live_captions.transcript",
+        name: "Live Captions & Transcripts",
+        domain: "live_captions",
+        category: CapabilityCategory::Automation,
+        description: "Real-time captioning from microphone or system audio with transcript \
+                      persistence and meeting-note summarization. Segments are streamed as \
+                      they arrive from the STT engine.",
+        how_to: "Start via RPC: openhuman.live_captions_start_transcript with source.",
+        status: CapabilityStatus::Beta,
+        privacy: Some(CapabilityPrivacy {
+            leaves_device: true,
+            data_kind: PrivacyDataKind::UserContent,
+            destinations: &["OpenHuman backend", "TinyHumans Neocortex"],
+        }),
+    },
+    // ── Voice Actions ────────────────────────────────────────────────────────
+    Capability {
+        id: "voice_actions.intent",
+        name: "Voice-Driven Actions",
+        domain: "voice_actions",
+        category: CapabilityCategory::Automation,
+        description: "Maps recognized utterances to controller-backed desktop actions with \
+                      safety levels (safe, requires_confirmation, destructive) and execution tracking.",
+        how_to: "Recognize via RPC: openhuman.voice_actions_recognize with utterance.",
+        status: CapabilityStatus::Beta,
+        privacy: Some(CapabilityPrivacy {
+            leaves_device: false,
+            data_kind: PrivacyDataKind::Derived,
+            destinations: &[],
+        }),
+    },
+    // ── Operator Inbox ──────────────────────────────────────────────────────
+    Capability {
+        id: "operator_inbox.triage",
+        name: "Operator Inbox Assistant",
+        domain: "operator_inbox",
+        category: CapabilityCategory::Automation,
+        description: "Channel-agnostic message triage with priority scoring, contextual draft \
+                      reply generation, and follow-up scheduling.",
+        how_to: "Triage via RPC: openhuman.operator_inbox_triage_message with sender/subject/body.",
+        status: CapabilityStatus::Beta,
+        privacy: Some(CapabilityPrivacy {
+            leaves_device: true,
+            data_kind: PrivacyDataKind::UserContent,
+            destinations: &["OpenHuman backend", "TinyHumans Neocortex"],
+        }),
+    },
+    // ── Chat with Data ──────────────────────────────────────────────────────
+    Capability {
+        id: "chat_with_data.query",
+        name: "Chat-with-Data Analytics",
+        domain: "chat_with_data",
+        category: CapabilityCategory::Automation,
+        description: "Natural-language querying over local/connected datasets with proactive \
+                      insight generation (anomaly detection, trend analysis, summaries).",
+        how_to: "Query via RPC: openhuman.chat_with_data_query with dataset_id and question.",
+        status: CapabilityStatus::Beta,
+        privacy: Some(CapabilityPrivacy {
+            leaves_device: true,
+            data_kind: PrivacyDataKind::UserContent,
+            destinations: &["OpenHuman backend", "TinyHumans Neocortex"],
+        }),
+    },
     // ── Mobile (iOS client) ─────────────────────────────────────────────────
     Capability {
         id: "mobile.device_pairing",

src/openhuman/about_app/types.rs

@@ -156,6 +156,8 @@ pub enum PrivacyDataKind {
     Diagnostics,
     /// Non-sensitive metadata (capability ids, feature flags, settings shape).
     Metadata,
+    /// User-generated content (audio, text, queries) sent to cloud LLM for inference.
+    UserContent,
 }
 
 #[cfg(test)]