AVADSA25 · AVADSA25 · Jul 28, 2026 · Jul 28, 2026 · Jul 28, 2026
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -2,9 +2,9 @@
 
 ## v3.5.0 (2026-07-22)
 ### Changed
-- **CODEC Pilot parked.** The browser-automation pillar is withdrawn from the product: the Pilot tab is removed from the dashboard and its product card from the Cortex map. Pilot needs a far deeper build than the rest of CODEC to be trustworthy — Google blocks account sign-in from any CDP-controlled browser (not a setting we can change), and cookie walls plus bot challenges make a large share of real sites unusable. The code is parked, not deleted: it lives in its own `codec-pilot` repo and the runner is untouched, so it can return when it earns its place.
+- **CODEC Pilot parked.** The browser-automation pillar is withdrawn from the product: the Pilot tab is removed from the dashboard and its product card from the Cortex map. Pilot needs a far deeper build than the rest of CODEC to be trustworthy — Google blocks account sign-in from any CDP-controlled browser (not a setting we can change), and cookie walls plus bot challenges make a large share of real sites unusable. The code is parked, not deleted: the skill, route proxy, and PM2 runner were unhooked from the repo and preserved in git history (see `docs/PILOT-PARKED.md`), so it can return when it earns its place.
 - **CODEC Project folded into CODEC Overview.** Project mode is a capability of the dashboard rather than a product in its own right. No functionality changes — plan → approve → autonomous run all work exactly as before.
-- **9 products → 7.** Core · Chat · Dashboard · Vibe · Agents · Dictate · Instant.
+- **9 products → 7.** Core · Dictate · Instant · Chat · Vibe · Voice · Overview.
 
 ### Added
 - **Claim-to-artifact matching (`codec_claim_check`).** CODEC may not claim what it did not do. Every real action is already recorded, so a claim of action with no corresponding action is false by construction — not by opinion. Catches both impossible capabilities ("I'll remember this for future sessions") and unbacked actions ("saved to your Desktop" with no file skill run). Covers the streaming and non-streaming chat paths. Tuned for false negatives over false positives: half the tests are false-positive guards.

diff --git a/DEMO_SCRIPT.md b/DEMO_SCRIPT.md
@@ -8,7 +8,7 @@
 > Status: ✅ verified · ⚠️ needs a decision/check before filming ·
 > 🔑 needs Google auth · 🧑 you run (voice/hardware).
 >
-> **Last sealed:** 2026-07-16.
+> **Last sealed:** 2026-07-28.
 
 ## Act 1 — It hears you, acts, and types for you
 
@@ -26,7 +26,7 @@
 | # | Feature | TRY (live) | EXPECT | Status |
 |---|---|---|---|---|
 | 7 | **Vision Mouse Control (UI-TARS)** — showstopper | On Cloudflare DNS: "Click the DNS button." | Cursor moves to the button and clicks — pixel coords, no accessibility API | ✅ |
-| 8 | Think mode + reveal reasoning | Ask a logic trap (rate / car-wash) with Think on → "Reveal train of thought" | Correct answer + readable reasoning panel | ✅ |
+| 8 | Think mode + reveal reasoning | Think on → "If 5 machines take 5 minutes to make 5 widgets, how long would 100 machines take to make 100 widgets?" → "Reveal train of thought" | Answers **"5 minutes"** (not 100) + a readable reasoning panel | ✅ |
 
 ## Act 3 — Autonomous agents (work while you watch)
 
@@ -71,37 +71,36 @@
 | 3-agent zero-dep concurrency | 9 | "Up to 3 background agents on a custom sub-800-line thread pool — no LangChain, no CrewAI." |
 | Plan-hash tamper + R/W grants | 10 | "The plan is sha256-hashed on approval — if the agent alters its own goals mid-run it auto-aborts. Reads and writes are glob-sandboxed to `~/codec-projects/`." |
 | ~~Self-healing replay + HITL takeover~~ (parked with Pilot) | — | "Your click is recorded as an element, not a pixel — so replay survives the page moving. It falls back XPath → CSS → local-LLM rescue; hit a captcha and you take the wheel, then hand back." |
-| CCF compression + temporal memory | 14 / 19 | "~65% token reduction; an FTS5 SQLite fact store with valid_from/valid_until — tell it your plans changed and it supersedes the old fact." |
-| Watchdog + blocked_on_qwen recovery | 14 / 15 | "A watchdog kills zombie >500MB/<0.5%-CPU processes; if the local model drops mid-project, the agent auto-resumes the moment the port recovers." |
+| CCF compression + temporal memory | 13 / 18 | "~65% token reduction; an FTS5 SQLite fact store with valid_from/valid_until — tell it your plans changed and it supersedes the old fact." |
+| Watchdog + blocked_on_qwen recovery | 13 / 14 | "A watchdog kills zombie >500MB/<0.5%-CPU processes; if the local model drops mid-project, the agent auto-resumes the moment the port recovers." |
 | AppKit overlay over fullscreen | any notification | "A native NSPanel floats status over *any* fullscreen app — watch it appear while I'm full-screen." |
-| Proactive nudges → iMessage/Telegram | 19 | "The observer can nudge you — a doc you've dwelled on — to a macOS banner, iMessage, or Telegram. The desktop agent reaches your pocket." |
+| Proactive nudges → iMessage/Telegram | 18 | "The observer can nudge you — a doc you've dwelled on — to a macOS banner, iMessage, or Telegram. The desktop agent reaches your pocket." |
 | Ed25519 signed self-updates | settings | "Updates verify an Ed25519-signed Sparkle appcast against an embedded key — hardened from execution all the way to updates." |
 
 ## Cut from the demo
 - **Compare across models** (ex-beat 21) — removed 2026-07-21: needs a licensed machine (AVA cloud) to show 3+ columns; on an unlicensed Mac it shows one and makes no comparison.
-- **CODEC Pilot** (ex-beat 13) — parked 2026-07-22. Google blocks account sign-in from any CDP-controlled browser, and cookie walls / bot challenges make a large share of real sites unusable. It needs a far deeper build to be trustworthy; the code is parked in the `codec-pilot` repo, not deleted.
+- **CODEC Pilot** (ex-beat 13) — parked 2026-07-22, unhooked from production in v3.5. Google blocks account sign-in from any CDP-controlled browser, and cookie walls / bot challenges make a large share of real sites unusable. It needs a far deeper build to be trustworthy; the code is preserved in git history (see `docs/PILOT-PARKED.md`), not deleted.
 - **Live webcam vision** (ex-beat 8) — removed 2026-07-16. The capture was soft (the Anker C200's lens, not code) and it landed as a gadget rather than a capability. Nothing else depends on it.
 
 ## Decide before filming
-- **Beat 18** — sign in to Notion once in the Connector tab (it persists now). Toggle **GitHub off** — it can't OAuth (needs a PAT).
-- Voice interrupt (now beat 21) — **confirmed working**.
+- **Beat 17** — sign in to Notion once in the Connector tab (it persists now). Toggle **GitHub off** — it can't OAuth (needs a PAT).
+- Voice interrupt (now beat 20) — **confirmed working**.
 
 ## Pre-record checklist
 - Chrome tabs L→R: `opencodec.org` → GitHub → Gmail → WhatsApp Web → Cloudflare → CODEC Chat
 - An empty text field / Notes open for the **F5 live-typing** beat (3)
 - Safari: Time Magazine article open
-- Claude Desktop: CODEC connector active **and set to Always Allow** (Act 5) — otherwise beat 17 stalls on a per-call approval prompt
+- Claude Desktop: CODEC connector active **and set to Always Allow** (Act 5) — otherwise beat 16 stalls on a per-call approval prompt
 - Phone: `codec.avadigital.ai` loaded, Touch ID ready
-- Notion signed in (Connector tab, beat 18); Philips Hue reachable; Spotify authorized (finale)
+- Notion signed in (Connector tab, beat 17); Philips Hue reachable; Spotify authorized (finale)
 - Vision Mouse tested 10× on the Cloudflare DNS button
 - Observer running with a populated buffer
-- **Pilot: do NOT rely on Google login** — Google blocks account sign-in from automated browsers. Beat 13 runs fully on public/anonymous pages (teach-mode + anonymous Gemini text). Skip Flow video-generation, which needs an account.
 - Cortex map open on a 2nd screen (persistent visual anchor)
 - Fresh Deep-Research + Project chat windows open
 
 ## Director's notes
 - Act 1 is the emotional hook — promise (1), deliver, and F5 (3) is an unexpected early "wow".
 - Showstopper risk: 7 (Vision Mouse) — if it fails live, skip to 8.
 - Core "works while you watch" pair: 10 Project, 12 Vibe.
-- Act 5 is THE differentiator: 17 (server) + 18 (client) back-to-back = bidirectional MCP.
-- Save Hue + Spotify (23) for the very last frame.
+- Act 5 is THE differentiator: 16 (server) + 17 (client) back-to-back = bidirectional MCP.
+- Save Hue + Spotify (22) for the very last frame.
diff --git a/FEATURES.md b/FEATURES.md
@@ -1,6 +1,6 @@
 # Sovereign AI Workstation — Full Product Breakdown
 
-> Engine: **CODEC v3.5** — 370 features · 90 skills · 2000+ tests · 52K+ lines of production code · 7 products
+> Engine: **CODEC v3.5** — 370 features · 89 skills · 2000+ tests · 52K+ lines of production code · 7 products
 
 The product name is **Sovereign AI Workstation**. Throughout this document
 and the codebase, **CODEC** refers to the underlying open-source engine /

diff --git a/README.md b/README.md
@@ -18,7 +18,7 @@
 <p align="center">
   <a href="https://github.com/AVADSA25/codec/actions/workflows/ci.yml"><img src="https://github.com/AVADSA25/codec/actions/workflows/ci.yml/badge.svg" alt="CI"/></a>
   <a href="https://github.com/AVADSA25/codec/discussions"><img src="https://img.shields.io/badge/community-Discussions-7057ff?style=flat-square&logo=github" alt="GitHub Discussions"/></a>
-  <a href="FEATURES.md"><img src="https://img.shields.io/badge/features-400+-blue?style=flat-square" alt="400+ Features"/></a>
+  <a href="FEATURES.md"><img src="https://img.shields.io/badge/features-370-blue?style=flat-square" alt="370 Features"/></a>
   <img src="https://img.shields.io/badge/skills-76-orange?style=flat-square" alt="76 Skills"/>
   <img src="https://img.shields.io/badge/tests-2000+-green?style=flat-square" alt="2,000+ Tests"/>
   <img src="https://img.shields.io/badge/lines-52K+-purple?style=flat-square" alt="52K+ Lines"/>
@@ -46,7 +46,7 @@ It listens, sees the screen, speaks back, controls apps, writes code, drafts mes
 
 No cloud dependency. No data leaving the machine unless you choose. No subscription on the open-source build. MIT licensed.
 
-> **Sovereign AI Workstation** is the product brand. **CODEC** (v3.2) is the open-source engine that powers it — the codename you'll see in code paths, skill registries, the `codec_*` PM2 services, and the `~/.codec/` config directory. *Sovereign AI Workstation* is what you ship; *CODEC* is what you ship with. Same way iPhone runs on Darwin, or Tesla Model S runs on Roadster components — one is the product, the other is the engine.
+> **Sovereign AI Workstation** is the product brand. **CODEC** (v3.5) is the open-source engine that powers it — the codename you'll see in code paths, skill registries, the `codec_*` PM2 services, and the `~/.codec/` config directory. *Sovereign AI Workstation* is what you ship; *CODEC* is what you ship with. Same way iPhone runs on Darwin, or Tesla Model S runs on Roadster components — one is the product, the other is the engine.
 
 ---
 
@@ -62,11 +62,11 @@ And all of it runs on *your* machine: no data leaves unless you explicitly route
 
 ---
 
-## 9 Products. One System.
+## 7 Products. One System.
 
 | # | Product | What It Does |
 |:-:|---|---|
-| 1 | **CODEC Core** | Voice command layer + vision mouse control — 75 skills, screen clicks by voice |
+| 1 | **CODEC Core** | Voice command layer + vision mouse control — 89 skills, screen clicks by voice |
 | 2 | **CODEC Dictate** | Hold, speak, paste — hands-free F5 live typing at cursor, draft refinement, floating overlays |
 | 3 | **CODEC Instant** | Right-click → 8 AI services system-wide — proofread, translate, reply, explain |
 | 4 | **CODEC Chat** | 250K-context conversational AI + 12 autonomous agent crews |
@@ -80,7 +80,7 @@ And all of it runs on *your* machine: no data leaves unless you explicitly route
 
 Always-on voice assistant. Say *"Hey CODEC"* or press F13 to activate. F18 for voice commands. F16 for text input.
 
-75 skills fire instantly: Google Calendar, Gmail, Drive, Docs, Sheets, Tasks, Keep, Chrome automation, web search, Hue lights, timers, Spotify, clipboard, terminal commands, PM2 control, and more. Most skills bypass the LLM entirely — direct action, zero latency. Skills are matched by trigger specificity — longer, more specific triggers always win over generic ones.
+89 skills fire instantly: Google Calendar, Gmail, Drive, Docs, Sheets, Tasks, Keep, Chrome automation, web search, Hue lights, timers, Spotify, clipboard, terminal commands, PM2 control, and more. Most skills bypass the LLM entirely — direct action, zero latency. Skills are matched by trigger specificity — longer, more specific triggers always win over generic ones.
 
 **Vision Mouse Control — See & Click**
 
@@ -114,7 +114,7 @@ Full conversational AI. Long context. File uploads (drag-and-drop). Image analys
 
 Voice input via continuous microphone with stop button. Streaming responses with typing and thinking indicators.
 
-**Four modes** in the chat composer: **Chat / Think / Agents / Project**. The first three live here — Project mode is documented as its own product in [§9 below](#9-codec-project--drop-a-project-autonomy) since it runs autonomously on a dedicated daemon for hours rather than turn-by-turn in the chat thread.
+**Four modes** in the chat composer: **Chat / Think / Agents / Project**. The first three live here — Project mode is documented within [§4 CODEC Chat](#4-codec-chat--250k-context--12-agent-crews--drop-a-project-autonomy) since it runs autonomously on a dedicated daemon for hours rather than turn-by-turn in the chat thread.
 
 Plus 12 autonomous agent crews — not single prompts, full multi-step workflows. Say *"research the latest AI agent frameworks and write a report."* Minutes later there's a formatted Google Doc in Drive with sources, images, and recommendations.
 
@@ -158,7 +158,7 @@ Full transcript saved to memory. Every conversation becomes searchable context f
 Private dashboard accessible from any device, anywhere. Cloudflare Tunnel or Tailscale VPN — no port forwarding, no third-party relay. 135+ API endpoints. Send commands, view the screen, launch voice calls, manage agents — all from a browser. Installable as a PWA on mobile and desktop.
 
 **Cortex — System Nerve Center**
-Visual command center showing all 9 CODEC products in an interactive grid. Neural network SVG map, real-time activity feed, searchable skills panel, and detailed event log viewer. The single-pane-of-glass view of the entire system.
+Visual command center showing all 7 CODEC products in an interactive grid. Neural network SVG map, real-time activity feed, searchable skills panel, and detailed event log viewer. The single-pane-of-glass view of the entire system.
 
 **Audit — Full Event Trail**
 Every action CODEC takes is logged across 16 categories: command, skill, llm, auth, error, scheduled, voice, vision, tts, stt, system, security, hotkey, screenshot, config, draft. Filterable by category pills, searchable, with colored timeline dots and expandable event details. JSON-line storage with 50MB rotation. Default 24h time range with 1h/6h/24h/7d quick filters.
@@ -282,7 +282,7 @@ Three smart agents ship built-in: Daily Briefing, Restaurant Decider (location-a
 </p>
 <p align="center">
   <img src="docs/screenshots/terminal.png" alt="Terminal" width="400"/><br/>
-  <em>75 skills loaded at startup</em>
+  <em>89 skills loaded at startup</em>
 </p>
 <p align="center">
   <img src="docs/screenshots/cortex.png" alt="Cortex Neural Map" width="720"/><br/>
@@ -703,7 +703,7 @@ setup_codec.py        — Setup wizard (9 steps)
 ecosystem.config.js   — PM2 process management (16 services)
 ```
 
-**CODEC Pilot** is parked (v3.5) and no longer part of the product. Its code and PM2 service live on in the `codec-pilot` repo.
+**CODEC Pilot** is parked (v3.5) and no longer part of the product. Its code, skill, route proxy, and PM2 service were unhooked from this repo in v3.5 and preserved in git history (see [docs/PILOT-PARKED.md](docs/PILOT-PARKED.md)).
 
 ---
 
@@ -743,7 +743,7 @@ python3 setup_codec.py
 
 > **Project mode** (drop-a-project autonomy) is live and ships as part of **CODEC Overview**.
 >
-> **CODEC Pilot** (browser automation) is **parked as of v3.5.** Google blocks account sign-in from any CDP-controlled browser, and cookie walls plus bot challenges make a large share of real sites unusable — it needs a considerably deeper build before it can be trusted as a product. The code is preserved in the `codec-pilot` repo rather than deleted.
+> **CODEC Pilot** (browser automation) is **parked as of v3.5.** Google blocks account sign-in from any CDP-controlled browser, and cookie walls plus bot challenges make a large share of real sites unusable — it needs a considerably deeper build before it can be trusted as a product. The code is preserved in git history (see [docs/PILOT-PARKED.md](docs/PILOT-PARKED.md)) rather than deleted.
 
 **Phase 3.5 (in progress)** — UX + polish on top of the autonomous-agent substrate: