ft-scout 9.0.3 → 9.0.5

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (84) hide show
  1. package/.firebase/hosting.d2Vi.cache +7 -7
  2. package/README.md +76 -83
  3. package/bin/src/commands/agent.d.ts.map +1 -1
  4. package/bin/src/commands/agent.js +31 -228
  5. package/bin/src/commands/agent.js.map +1 -1
  6. package/bin/src/commands/plan.d.ts.map +1 -1
  7. package/bin/src/commands/plan.js +45 -18
  8. package/bin/src/commands/plan.js.map +1 -1
  9. package/bin/src/engine/actionVerifier.d.ts.map +1 -1
  10. package/bin/src/engine/actionVerifier.js +61 -9
  11. package/bin/src/engine/actionVerifier.js.map +1 -1
  12. package/bin/src/engine/agentEngine.d.ts.map +1 -1
  13. package/bin/src/engine/agentEngine.js +701 -553
  14. package/bin/src/engine/agentEngine.js.map +1 -1
  15. package/bin/src/engine/agentState.d.ts.map +1 -1
  16. package/bin/src/engine/agentState.js +0 -64
  17. package/bin/src/engine/agentState.js.map +1 -1
  18. package/bin/src/engine/browser/browserController.d.ts.map +1 -1
  19. package/bin/src/engine/browser/browserController.js +8 -0
  20. package/bin/src/engine/browser/browserController.js.map +1 -1
  21. package/bin/src/engine/computer/computerController.d.ts.map +1 -1
  22. package/bin/src/engine/computer/computerController.js +686 -240
  23. package/bin/src/engine/computer/computerController.js.map +1 -1
  24. package/bin/src/engine/computer/computerDiagnostic.d.ts.map +1 -0
  25. package/bin/src/engine/computer/computerDiagnostic.js +71 -0
  26. package/bin/src/engine/computer/computerDiagnostic.js.map +1 -0
  27. package/bin/src/engine/computer/index.d.ts.map +1 -1
  28. package/bin/src/engine/computer/index.js +2 -0
  29. package/bin/src/engine/computer/index.js.map +1 -1
  30. package/bin/src/engine/computer/mockComputer.d.ts.map +1 -0
  31. package/bin/src/engine/computer/mockComputer.js +260 -0
  32. package/bin/src/engine/computer/mockComputer.js.map +1 -0
  33. package/bin/src/engine/computer/scoutDriver.cs +873 -0
  34. package/bin/src/engine/computer/scout_driver.exe +0 -0
  35. package/bin/src/engine/runtime/agentRuntime.d.ts.map +1 -0
  36. package/bin/src/engine/runtime/agentRuntime.js +1160 -0
  37. package/bin/src/engine/runtime/agentRuntime.js.map +1 -0
  38. package/bin/src/engine/runtime/index.d.ts.map +1 -0
  39. package/bin/src/engine/runtime/index.js +2 -0
  40. package/bin/src/engine/runtime/index.js.map +1 -0
  41. package/bin/src/engine/safety.d.ts.map +1 -1
  42. package/bin/src/engine/safety.js +59 -25
  43. package/bin/src/engine/safety.js.map +1 -1
  44. package/bin/src/engine/toolRegistry.d.ts.map +1 -1
  45. package/bin/src/engine/toolRegistry.js +293 -68
  46. package/bin/src/engine/toolRegistry.js.map +1 -1
  47. package/bin/src/engine/voiceEngine.d.ts.map +1 -1
  48. package/bin/src/engine/voiceEngine.js +0 -1
  49. package/bin/src/engine/voiceEngine.js.map +1 -1
  50. package/bin/src/engine/workspace/agentWorkspace.d.ts.map +1 -0
  51. package/bin/src/engine/workspace/agentWorkspace.js +505 -0
  52. package/bin/src/engine/workspace/agentWorkspace.js.map +1 -0
  53. package/bin/src/engine/workspace/index.d.ts.map +1 -0
  54. package/bin/src/engine/workspace/index.js +2 -0
  55. package/bin/src/engine/workspace/index.js.map +1 -0
  56. package/bin/src/index.js +4 -5
  57. package/bin/src/index.js.map +1 -1
  58. package/bin/src/server/server.d.ts.map +1 -1
  59. package/bin/src/server/server.js +25 -4
  60. package/bin/src/server/server.js.map +1 -1
  61. package/bin/src/utils/auth.d.ts.map +1 -1
  62. package/bin/src/utils/auth.js +81 -50
  63. package/bin/src/utils/auth.js.map +1 -1
  64. package/bin/src/utils/branding.js +1 -1
  65. package/bin/src/utils/payment.d.ts.map +1 -1
  66. package/bin/src/utils/payment.js +15 -1
  67. package/bin/src/utils/payment.js.map +1 -1
  68. package/bin/src/utils/pricing.js +1 -1
  69. package/bin/src/utils/pricing.js.map +1 -1
  70. package/bin/src/utils/razorpay.d.ts.map +1 -1
  71. package/bin/src/utils/razorpay.js +10 -2
  72. package/bin/src/utils/razorpay.js.map +1 -1
  73. package/downloads/invoice-2026-03.pdf +1 -0
  74. package/package.json +1 -1
  75. package/web/app.js +54 -135
  76. package/web/styles.css +2 -2
  77. package/bin/src/engine/appControl.d.ts.map +0 -1
  78. package/bin/src/engine/appControl.js +0 -2413
  79. package/bin/src/engine/appControl.js.map +0 -1
  80. package/bin/src/engine/liveScreenEngine.d.ts.map +0 -1
  81. package/bin/src/engine/liveScreenEngine.js +0 -421
  82. package/bin/src/engine/liveScreenEngine.js.map +0 -1
  83. package/bin/src/engine/scripts/winCapture.ps1 +0 -248
  84. package/downloads/INVOICE_2026_ACME.pdf +0 -1
@@ -1,8 +1,8 @@
1
+ web_cli.html,1789281384143,d251153dac472e4e4ea21d00f881b575e81a63822da13006e720c376cfbe24af
2
+ styles.css,1789281384145,f1c7c32071cbb851d4f0a3023375765c3a5630106878c5fcad8456adb98e138a
1
3
  ScoutFavicon.png,1786186569805,41a9ca9abf7acd2976e78c72a72285b49cc0e64b2b5eeb2c2b88c69628fc618a
2
- paid_plans.html,1789149791483,fc230a3ca475280bd09cd608d561fa82c96224a006c3525618cc96c148ab8415
3
- styles.css,1789148412834,88bbfea5905ed53d1ccbcd026f5f5bec4a1b4f7df75f33d4d2623d1ea7704b7d
4
- authentication.html,1789148412837,264926efc21b4099e0e3883c0ddc33d3c2bc3dfd303338660b10c1d9566b7563
5
- index.html,1789148412836,0a93595b018c389e773c76767142a40a4bcd1dad6ee3a97e9844e8d4a7d80bc2
6
- web_cli.html,1789148412832,3670d937ef638d6806511bb5b92ca24143044ad8675a3115d2e0292ec96243f6
7
- admin.html,1789149713352,5b311424b315f8b9c0663f76b5c5040b00f49153fb90748fd07101e784e8fab6
8
- app.js,1789150141943,d7e37c935df1ced6392cf46df90cc50e341ef0f4c1c09c42464f55ae91660af0
4
+ authentication.html,1789281384176,678dbfb5438ff13b0a6f8d759c1af77d5aac926ff44ea7815f181e80674d183a
5
+ app.js,1789281384176,f281ff2c345623a652af75a4db565bb35377bfd452a5e5af940c362c65d2f52f
6
+ admin.html,1789281384176,72bb3b4ab7dbbe6ff32bfb5a51e8c79ef0bece4e2f1fc91218296df17e9b34f4
7
+ paid_plans.html,1789283167535,0708329edc4938577250c92fb1a294d09ea38778b8edec990b0b80adc839e49f
8
+ index.html,1789282851111,f7812b28faad4cc0ebf3062272dfadc297ab257d3d73ff254b69780e5401934a
package/README.md CHANGED
@@ -1,4 +1,4 @@
1
- # scout (FrontTerrain CLI v9.0.3)
1
+ # scout (FrontTerrain CLI v9.0.5)
2
2
 
3
3
  **FrontTerrain (`scout`)** — The Autonomous AI Principal Software Engineer & Codebase Co-Pilot in your terminal.
4
4
 
@@ -20,28 +20,24 @@ npm install -g ft-scout@latest
20
20
 
21
21
  Once installed, use `scout` or `ft` in any project directory:
22
22
 
23
- ### Autonomous AI Agent & Interactive Sessions
23
+ ### Autonomous Computer-Use Operator & AI Agent Sessions
24
24
 
25
25
  ```bash
26
26
  # Autonomous Multi-Step AI Agent (Direct goal execution, context inspection, automated test/build verification)
27
+ scout agent "Open Notepad, type 'Autonomous test note', and save to Desktop"
27
28
  scout agent "Build a responsive snake game in HTML and JS with smooth animations"
28
- scout agent "Add persistent score tracking and level progression"
29
29
  scout task "Fix typescript build errors in src/utils" -v "npm run build"
30
30
  scout goal "Refactor database query logic" -y
31
31
 
32
- ### Interactive Scout Agent Session (Persistent chat loop with Slash Commands & full ReAct tool execution)
32
+ ### Interactive Scout Agent Session (Persistent operator loop with Slash Commands & full tool execution)
33
33
  scout agent
34
34
 
35
- # Select Interaction Mode: Text Interaction | Audio Interaction (Voice STT + TTS) | Live Screen Share Interaction
36
- # Launch directly into your preferred mode via CLI flags:
37
- scout agent -S # Launch with Live Screen Sharing (streams active app window to model in real time)
38
- scout agent --screen-share
35
+ # Optional voice interaction flags:
39
36
  scout agent -V # Launch with Audio Voice mode (STT + TTS)
40
37
  scout agent --voice
41
38
 
42
39
  # Inside session:
43
- # Agent Prompt: "Inspect Chrome and automate login form submission"
44
- # Slash commands: /help /screenshare /scratchpad /voice /speak /tree /stats /test /explain /export /diff /undo /clear /history /exit
40
+ # Slash commands: /help /voice /speak /tree /stats /test /explain /export /diff /undo /clear /history /exit
45
41
 
46
42
  # Generate and create a new source code file
47
43
  scout agent --create src/utils/math.ts "Create helper functions for sum, average, and array chunking"
@@ -141,7 +137,7 @@ scout benchmark
141
137
 
142
138
  ## Autonomous Computer-Use Agent Architecture
143
139
 
144
- Scout is upgraded into a genuinely autonomous computer-use agent capable of executing complex multi-step workflows across desktop environments, browsers, and terminal sessions:
140
+ Scout is built as a high-agency autonomous operator capable of operating external desktop applications, browsers, operating systems, and terminal workspaces while strictly preserving user intent:
145
141
 
146
142
  ```
147
143
  ┌────────────────────────────────────────────────────────┐
@@ -155,9 +151,18 @@ Scout is upgraded into a genuinely autonomous computer-use agent capable of exec
155
151
 
156
152
 
157
153
  ┌────────────────────────────────────────────────────────┐
154
+ │ Execution Intent Engine │
155
+ │ • Inferred Intent: Computer / Browser / Code / Mixed │
156
+ │ • Required Apps: [Notepad, Excel, Chrome, ...] │
157
+ │ • Destination: Desktop / Downloads / Repository Root │
158
+ │ • Constraint: USER INTENT > TOOL CONVENIENCE │
159
+ └──────────────────────────┬─────────────────────────────┘
160
+
161
+
162
+ ┌────────────────────────────────────────────────────────┐
158
163
  │ Task State Manager │
159
164
  │ • Checkpoints: .ft/tasks/<taskId>.json │
160
- │ • Working Memory: .ft/scratchpad.md
165
+ │ • Structured Memory: AgentWorkspace & Intent Gates
161
166
  │ • Resume Support: scout agent --resume <taskId> │
162
167
  └──────────────────────────┬─────────────────────────────┘
163
168
 
@@ -172,53 +177,71 @@ Scout is upgraded into a genuinely autonomous computer-use agent capable of exec
172
177
  └─────────────────────┬─────────────────────┘
173
178
 
174
179
 
175
- ┌────────────────────┐
176
- Tool Execution
177
- │ • Computer Control
178
- Browser Engine
179
- │ • File / CLI Ops
180
- └──────────┬─────────┘
180
+ ┌────────────────────────────────────────────────────────┐
181
+ Tool Execution Guard
182
+ │ • Strictly blocks direct filesystem shortcuts when
183
+ application interaction was requested
184
+ │ • Native Computer Controller (PowerShell, AppleScript)
185
+ │ • Semantic Browser Engine & Core Codebase Tools │
186
+ └──────────────────────────┬─────────────────────────────┘
181
187
 
182
188
 
183
189
  ┌────────────────────┐
184
190
  │ Action Verifier │
185
- │ • File checks
186
- │ • DOM validation
187
- │ • URL matches
191
+ │ • File existence
192
+ │ • Desktop vs repo ◄── Rejects substitute paths
193
+ │ • Window active
194
+ │ • Content check │
188
195
  └──────────┬─────────┘
189
196
 
190
197
  ┌─────────────────────┴─────────────────────┐
191
- │ Success │ Failure
198
+ │ Success │ Failure (Structured Error)
192
199
  ▼ ▼
193
200
  ┌─────────────────────┐ ┌─────────────────────┐
194
- │ Next Step / │ │ Error Recovery
195
- │ Completion │ │ • RETRY
196
- └─────────────────────┘ │ • ALTERNATIVE
197
- │ • ROLLBACK / ABORT
201
+ │ Next Step / │ │ Bounded Recovery
202
+ │ Completion │ │ • OBSERVE (capture)
203
+ REFOCUS WINDOW
204
+ └─────────────────────┘ │ • RETRY COMPUTER OP
198
205
  └─────────────────────┘
199
206
  ```
200
207
 
201
- ### 1. Task State Persistence & Resumability
202
- - **Dual State Architecture**: Tracks goal metadata, step progression, executed actions, modified files, and checkpoints in `.ft/tasks/<taskId>.json`.
203
- - **Live Markdown Scratchpad**: Generates and updates `.ft/scratchpad.md` in real-time with human-readable checklists, active memory, and plan state.
204
- - **Task Resumption**: Re-attach to any incomplete or paused task with `scout agent --resume <taskId>` without starting from scratch.
208
+ ### 1. Execution Intent Preservation (`USER INTENT > TOOL CONVENIENCE`)
209
+ - **Outcome vs. Execution Requirements**: Scout clearly differentiates between outcome tasks (*"Create file X"*) and execution tasks (*"Open Notepad, type X, and save to Desktop"*).
210
+ - **Anti-Substitution Gating**: When a user specifies desktop application interaction, Scout strictly prevents silent shortcuts (such as invoking direct filesystem APIs) and executes the requested workflow through the application interface.
211
+ - **True Desktop Path Resolution**: Automatically discovers and targets the user's actual Desktop path (including Windows OneDrive/Desktop and `~/Desktop`), rejecting accidental substitutions into repository roots.
205
212
 
206
- ### 2. Risk Classification & Safety Guardrails
207
- - **Action-Risk Classification**: Evaluates each tool action before execution into `LOW`, `MEDIUM`, `HIGH`, or `CRITICAL` risk tiers.
208
- - **Human-in-the-Loop Approvals**: Automatically pauses and queues sensitive operations (file deletions, system settings, external communications, high-impact terminal commands) for explicit human approval via the `ApprovalManager`.
213
+ ### 2. Native Computer & Desktop Application Control
214
+ - **Cross-Platform System Controller**: Native mouse movement, clicking (`left`, `right`, `double`), text typing (`SendWait`), keyboard combinations (`Ctrl+S`, `Alt+F4`), window activation, and UI inspection across Windows (PowerShell `-EncodedCommand`), macOS (AppleScript), and Linux (`xdotool`).
215
+ - **Active Application State Tracking**: Continuously monitors `expectedApplication`, `activeApplication`, and `activeWindow` before and after UI actions.
216
+ - **Bounded Error Recovery Loop**: When an action fails (e.g. lost foreground focus), Scout returns structured feedback `{ success: false, action, error, recoverable: true }` and executes a bounded recovery cycle (`OBSERVE ➔ REFOCUS ➔ RETRY`) instead of abandoning the interface.
209
217
 
210
- ### 3. Action Verifier & Error Recovery Engine
211
- - **Deterministic Post-Condition Checks**: Verifies actual outcome rather than assuming success (`file_exists`, `file_syntax`, `dom_element_present`, `url_matches`, `file_content_matches`).
212
- - **Dynamic Recovery Strategies**: Automatically selects between `RETRY`, `ALTERNATIVE` tool actions, `ROLLBACK`, or graceful `ABORT` with full diagnostic reporting.
218
+ ### 3. Task State Persistence & Resumability
219
+ - **Checkpointed Task Architecture**: Real-time step progress, observations, artifacts, and execution intents are continuously persisted to `.ft/tasks/<taskId>.json`.
220
+ - **Operational Memory Compaction**: Context summarization prevents unbounded transcript growth during long multi-step workflows while keeping active goals and variables visible to the model.
221
+ - **Task Resumption**: Re-attach to any paused or interrupted session with `scout agent --resume <taskId>`.
213
222
 
214
- ### 4. Computer & Browser Automation
215
- - **Desktop Computer Controller**: Mouse click coordination (`left`, `right`, `double`, `middle`, `scroll`), smooth cursor positioning, text typing, keystroke injection, and active window management.
216
- - **Deterministic Browser Engine**: URL navigation, DOM hierarchy inspection, form input submission, button clicks, and download tracking across web sessions.
223
+ ### 4. Safety Guardrails & Human-in-the-Loop Approvals
224
+ - **Risk Classification**: Classifies every action as `LOW`, `MEDIUM`, `HIGH`, or `CRITICAL`.
225
+ - **Approval Gate**: Pauses and queues high-impact operations (destructive file deletes, system commands, credential requests) for explicit human authorization.
217
226
 
218
227
  ---
219
228
 
220
- ## Scout Autonomous ReAct Tools (18 Tools Available)
229
+ ## Scout Autonomous Tool Registry
221
230
 
231
+ ### Computer & Application Control Tools
232
+ | Tool | Description |
233
+ |---|---|
234
+ | `computer_screenshot` | Capture display screen image and inspect active foreground window |
235
+ | `computer_click` | Move cursor and click mouse at coordinate (left, right, middle) |
236
+ | `computer_double_click` | Double-click mouse at specified screen coordinate |
237
+ | `computer_type` | Ensure target window focus and type text keystrokes via OS keyboard |
238
+ | `computer_key_press` | Press individual keyboard key (`enter`, `tab`, `escape`, `backspace`) |
239
+ | `computer_hotkey` | Trigger key combination (`["ctrl", "s"]`, `["alt", "tab"]`) |
240
+ | `computer_open_app` | Launch desktop application (Notepad, Excel, Chrome, Terminal) and focus window |
241
+ | `computer_switch_window` | Switch foreground active window to target application |
242
+ | `computer_inspect_ui` | Inspect active window coordinates, title, and UI state |
243
+
244
+ ### Workspace & Engineering Tools
222
245
  | Tool | Description |
223
246
  |---|---|
224
247
  | `read_file` | Read content & line ranges of workspace files |
@@ -234,23 +257,8 @@ Scout is upgraded into a genuinely autonomous computer-use agent capable of exec
234
257
  | `fetch_url` | Fetch web API documentation, specs, or external web URLs |
235
258
  | `git_diff` | Query uncommitted workspace git diff directly inside execution loop |
236
259
  | `ask_user` | Request missing secret credentials or API keys |
237
- | `open_app` | Launch or focus external applications (browser, terminal, editor, system apps) |
238
- | `app_action` | Desktop takeover, cursor movement, mouse clicks, keystrokes, and window focus |
239
- | `agent_scratchpad` | Update, read, or clear agent working memory, step checklist, & notes (`.ft/scratchpad.md`) |
240
260
  | `speak_text` | Read text phrases out loud using Text-to-Speech (TTS) voice synthesis |
241
- | `task_completed` | Conclude autonomous goal execution with structured teardown report |
242
-
243
- ---
244
-
245
- ## Desktop Automation & External App Takeover
246
-
247
- Scout can interact with and automate desktop applications, web browsers, and system tools outside of the terminal:
248
-
249
- - **Full Application Takeover (`takeover`)**: Focuses target windows, brings them to the foreground, positions the cursor, and executes multi-step automation workflows.
250
- - **Cursor & Mouse Control (`click_app`, `move_mouse`)**: Cross-platform mouse positioning and clicking with support for `left`, `right`, `double`, `middle`, and `scroll` actions.
251
- - **Keystroke & Hotkey Injection (`type_text`, `send_keys`)**: Direct text typing with auto-enter, plus keyboard shortcuts (`Ctrl+C`, `Ctrl+V`, `Cmd+Option+I`, etc.).
252
- - **Zero-Screenshot Privacy Architecture**: Operates with zero desktop screenshot capture or storage. Automation relies on deterministic window focusing, coordinate clicks, and direct keystrokes/hotkeys for 100% desktop privacy and zero latency.
253
- - **Persistent Scratchpad Memory (`agent_scratchpad`)**: Maintains live planning notes, step-by-step checklists, and execution status inside `.ft/scratchpad.md` across complex multi-step sessions.
261
+ | `task_completed` | Conclude autonomous goal execution with verification and report |
254
262
 
255
263
  ---
256
264
 
@@ -259,29 +267,17 @@ Scout can interact with and automate desktop applications, web browsers, and sys
259
267
  Scout is built for uninterrupted autonomy across multiple LLM providers:
260
268
 
261
269
  - **Multi-Provider Routing**: Seamlessly routes and fails over across **NVIDIA NIM** (Nemotron, DeepSeek R1/V3), **Groq** (Llama 3.3 70B), **Google Gemini**, and **OpenAI** (GPT-4o).
262
- - **Universal Tool-Calling Fallback**: For reasoning models or providers that do not natively support OpenAI tool/function calling (e.g., Nemotron, DeepSeek reasoning models), Scout automatically injects structured JSON schema instructions and robustly parses tool invocations. Tools like `app_action`, `open_app`, and `agent_scratchpad` trigger reliably on every model.
270
+ - **Universal Tool-Calling Fallback**: For reasoning models or providers that do not natively support OpenAI tool/function calling (e.g., Nemotron, DeepSeek reasoning models), Scout automatically injects structured JSON schema instructions and robustly parses tool invocations.
263
271
  - **Adaptive Timeout Handling**: Extended 90-second execution window for deep reasoning models on NVIDIA NIM.
264
272
  - **Instant Model Failover**: Automatically detects timeouts (`ETIMEDOUT`, `ECONNABORTED`, `408`, `AbortError`), payload errors, and tool-definition errors, instantly switching to fallback models without wasting retry cycles.
265
273
 
266
274
  ---
267
275
 
268
- ## Triple Interaction Modes (Text, Voice, & Live Screen Share)
276
+ ## Dual Interaction Modes (Text & Voice)
269
277
 
270
- When starting interactive sessions in `scout agent`, `scout init` (ask mode), or `scout search`, Scout presents a **3-Option Interaction Selector**:
278
+ When starting interactive sessions in `scout agent`, `scout init` (ask mode), or `scout search`, Scout presents interactive options:
271
279
  1. **Text Interaction**: Standard command/prompt input -> Text response in terminal.
272
280
  2. **Audio Interaction**: Voice command input via microphone Speech-to-Text (STT) -> Text response + Audio response spoken out loud via Text-to-Speech (TTS).
273
- 3. **Live Screen Share Interaction**: Real-time live screen share stream directly to AI vision models + Target App Window Isolation + Continuous Futuristic Sky-Blue Aura HUD.
274
-
275
- ---
276
-
277
- ## Real-Time Live Screen Sharing & Visual Perception Engine
278
-
279
- Scout features a native in-memory live screen sharing engine designed specifically for multimodal AI co-pilot workflows:
280
-
281
- - **Direct In-Memory Streaming (0 Files Saved to Disk)**: Completely eliminates saving temporary `.png` screenshot files to disk. Real-time compressed frames stream directly in-memory to AI multimodal vision models (Google Gemini 2.0 Flash, NVIDIA Vision, OpenAI GPT-4o) with zero disk I/O overhead.
282
- - **Target Application Window Isolation**: When the agent is operating on an external app (such as Google Chrome, Microsoft Edge, VS Code, or Notepad), the capture engine isolates **ONLY that specific application's window rectangle** (`BoundingRectangle`). Background applications, sensitive tabs, and other monitor areas are completely excluded from the stream.
283
- - **Continuous Sky-Blue Aura HUD Overlay**: While live screen sharing is active, a glowing sky-blue (`#00e5ff`) aura border and corner HUD pulses smoothly across screen edges with an active status badge (`Live Screen Share Active [App: chrome] • Streaming to Model Perception Active`) rendered with UTF-8 BOM and XML entity encoding for crisp typography.
284
- - **Autonomous App Focus Synchronization**: Whenever the agent opens or switches applications (`openApp('chrome')` or `takeoverControl('chrome')`), the live stream viewport automatically retargets and locks onto that application window.
285
281
 
286
282
  ---
287
283
 
@@ -289,9 +285,7 @@ Scout features a native in-memory live screen sharing engine designed specifical
289
285
 
290
286
  Inside an active `scout agent` interactive session, type any slash command to trigger agent actions:
291
287
 
292
- - `/help` — Display comprehensive guide to all 18 tools and interactive slash commands.
293
- - `/screenshare` or `/livescreen` — Toggle real-time in-memory screen share stream to the AI model.
294
- - `/scratchpad` — View active working memory plan & step checklist (`.ft/scratchpad.md`).
288
+ - `/help` — Display comprehensive guide to all 15 tools and interactive slash commands.
295
289
  - `/voice` — Activate microphone Speech-to-Text (STT) listener to speak your prompt into CLI.
296
290
  - `/speak <text>` — Read entered text out loud via Text-to-Speech (TTS) voice synthesis.
297
291
  - `/tree [dir]` — Render an interactive visual directory tree of any workspace folder.
@@ -307,28 +301,27 @@ Inside an active `scout agent` interactive session, type any slash command to tr
307
301
 
308
302
  ---
309
303
 
310
- ## What's New in v9.0.3
311
- - **Autonomous Computer-Use Agent Upgrade**: Built-in state management with task checkpoints (`.ft/tasks/`), plan decomposition, risk classification, approval gates, action verification, and self-healing recovery loops.
304
+ ## What's New in v9.0.5
305
+ - **Autonomous Computer-Use & Desktop App Operator**: Native desktop application control (Notepad, Excel, VS Code, Browser) with full mouse, keyboard, hotkey, and window focus management across Windows, macOS, and Linux.
306
+ - **Strict Execution Intent Alignment (`USER INTENT > TOOL CONVENIENCE`)**: Differentiates between outcome and execution requirements; strictly blocks silent filesystem shortcuts when application interaction is requested.
307
+ - **Bounded Computer Recovery Loop**: Structured action feedback (`{ success, action, error, recoverable }`) with automated re-observation, window refocusing, and retries.
308
+ - **Action Verifier & Anti-Substitute Path Rejection**: Verifies true destination (resolving Windows OneDrive/Desktop and `~/Desktop`) and rejects substitute files created in the workspace root.
312
309
  - **Interactive Web Dashboard & Sync Server (`scout dashboard`)**: Launch a local web GUI to visually inspect repository maps, audit logs, plan details, and live agent controls.
313
310
  - **Developer Authentication Guard (`scout login` / `signup` / `auth`)**: Unified account management with Firebase and Holosene token authentication, securing command execution and session persistence.
314
311
  - **Quota & Subscription Management (`scout quota` / `plan`)**: Easily inspect real-time usage quotas, track billing renewal cycles, switch between tiers, or configure BYOK (Bring-Your-Own-Key) access.
315
312
  - **Automated Model Benchmark Suite (`scout benchmark`)**: Run live end-to-end latency and throughput benchmarks (tokens/sec) across all configured LLM providers (Nemotron, Qwen, Gemini, GPT).
316
- - **Real-Time Live Screen Sharing Engine**: In-memory video streaming directly to vision models with zero disk writes.
317
- - **Target App Window Isolation**: Selectively captures and streams ONLY the active application window the agent is working in (e.g., Chrome).
318
- - **Continuous Sky-Blue Aura HUD**: Pulsing `#00e5ff` HUD border and corner brackets with clean UTF-8 bullet rendering.
319
313
  - **Nemotron & DeepSeek Fallback Tool-Calling**: Smart JSON prompt formatting and fallback parser for models lacking native function-calling.
320
314
  - **Extended 90s LLM Timeout & Instant Failover**: Extended execution window for deep reasoning models and immediate failover upon timeout or connection abort.
321
- - **Desktop Takeover & Scratchpad Tasks**: Cross-platform cursor positioning, mouse clicking, keystroke injection, and external app automation.
322
315
 
323
316
  ---
324
317
 
325
318
  ## Overhauled Features & Performance
326
319
 
327
- - **Autonomous Multi-Step Computer-Use**: Decompose complex tasks into structured steps, execute in browsers and desktop apps, verify outcomes, and self-heal from failures.
328
- - **Task State Persistence (`.ft/tasks/` & `.ft/scratchpad.md`)**: Maintains checkpointed task states and scratchpad checklists across execution sessions.
320
+ - **Autonomous Multi-Step Computer-Use**: Decompose complex tasks into structured steps, verify outcomes, and self-heal from failures.
321
+ - **Strict Intent Governance**: Preserves required applications and target paths without unauthorized tool substitutions.
322
+ - **Task State Persistence (`.ft/tasks/`)**: Maintains checkpointed task states across execution sessions.
329
323
  - **Safety & Human Approval**: Automatic risk rating with human-in-the-loop intervention for high-risk operations.
330
324
  - **In-Terminal Native STT & TTS Voice Engine**: Direct in-terminal microphone voice input and OS native speech synthesis with zero browser popups.
331
- - **Desktop Automation & App Takeover (`open_app`, `app_action`)**: Seamlessly launch, navigate, take over focus, control cursor, and send keystrokes across external applications.
332
325
  - **Multi-Model Provider Resilience**: Intelligent automatic routing and failover across NVIDIA NIM, Groq, Gemini, and OpenAI with universal JSON tool-call fallback.
333
326
  - **High-Agency Multi-File Scaffolding**: Automatically creates, implements, tests, and verifies multi-file codebases from scratch.
334
327
  - **Atomic Multi-Snippet Editing (`multi_edit_file`)**: Executes multiple code replacements in a single turn for complex multi-point refactoring.
@@ -1 +1 @@
1
- {"version":3,"file":"agent.d.ts","sourceRoot":"","sources":["agent.ts"],"names":[],"mappings":"AA+BA,MAAM,WAAW,YAAY;IACzB,EAAE,EAAE,MAAM,CAAC;IACX,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,MAAM,CAAC;IACb,QAAQ,EAAE,MAAM,GAAG,QAAQ,GAAG,KAAK,CAAC;IACpC,WAAW,EAAE,MAAM,CAAC;CACvB;AAKD,wBAAgB,cAAc,CAAC,MAAM,EAAE,YAAY,EAAE,QAKpD;AAED,wBAAgB,cAAc,IAAI,YAAY,EAAE,CAO/C;AAED,MAAM,WAAW,YAAY;IACzB,IAAI,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC1B,MAAM,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC5B,IAAI,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC1B,KAAK,CAAC,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAAC;IACpC,GAAG,CAAC,EAAE,MAAM,GAAG,MAAM,EAAE,GAAG,SAAS,CAAC;IACpC,IAAI,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IAC3B,GAAG,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IACzB,IAAI,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC1B,SAAS,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC/B,UAAU,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAChC,KAAK,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC3B,QAAQ,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC9B,IAAI,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC1B,MAAM,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IAC7B,GAAG,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IAC1B,KAAK,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IAC5B,WAAW,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IAClC,MAAM,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;CAC/B;AAED,wBAAgB,wBAAwB,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,EAAE,CAsDlE;AAED,wBAAgB,uBAAuB,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAG3E;AAED,MAAM,WAAW,gBAAgB;IAC7B,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,UAAU,GAAG,QAAQ,GAAG,QAAQ,GAAG,MAAM,GAAG,MAAM,GAAG,KAAK,GAAG,KAAK,CAAC;IAC/E,UAAU,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAChC,WAAW,CAAC,EAAE,MAAM,EAAE,GAAG,SAAS,CAAC;CACtC;AAED,wBAAgB,qBAAqB,CAAC,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,GAAG,gBAAgB,CA2F5F;AAED,wBAAgB,yBAAyB,CACrC,UAAU,EAAE,MAAM,EAClB,aAAa,EAAE,MAAM,GAAG,SAAS,EACjC,SAAS,EAAE,MAAM,EAAE,GACpB,MAAM,EAAE,CAgGV;AAED,wBAAsB,iBAAiB,CAAC,cAAc,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,GAAG,MAAM,CAAC,CAuLxF;AAED,wBAAsB,WAAW,CAAC,SAAS,CAAC,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE,YAAY,iBAkpC3E"}
1
+ {"version":3,"file":"agent.d.ts","sourceRoot":"","sources":["agent.ts"],"names":[],"mappings":"AA6BA,MAAM,WAAW,YAAY;IACzB,EAAE,EAAE,MAAM,CAAC;IACX,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,MAAM,CAAC;IACb,QAAQ,EAAE,MAAM,GAAG,QAAQ,GAAG,KAAK,CAAC;IACpC,WAAW,EAAE,MAAM,CAAC;CACvB;AAKD,wBAAgB,cAAc,CAAC,MAAM,EAAE,YAAY,EAAE,QAKpD;AAED,wBAAgB,cAAc,IAAI,YAAY,EAAE,CAO/C;AAED,MAAM,WAAW,YAAY;IACzB,IAAI,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC1B,MAAM,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC5B,IAAI,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC1B,KAAK,CAAC,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAAC;IACpC,GAAG,CAAC,EAAE,MAAM,GAAG,MAAM,EAAE,GAAG,SAAS,CAAC;IACpC,IAAI,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IAC3B,GAAG,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IACzB,IAAI,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC1B,SAAS,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC/B,UAAU,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAChC,KAAK,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC3B,QAAQ,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC9B,IAAI,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC1B,MAAM,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IAC7B,GAAG,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IAC1B,KAAK,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IAC5B,MAAM,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;CAC/B;AAED,wBAAgB,wBAAwB,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,EAAE,CAsDlE;AAED,wBAAgB,uBAAuB,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAG3E;AAED,MAAM,WAAW,gBAAgB;IAC7B,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,UAAU,GAAG,QAAQ,GAAG,QAAQ,GAAG,MAAM,GAAG,MAAM,GAAG,KAAK,CAAC;IACvE,UAAU,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAChC,WAAW,CAAC,EAAE,MAAM,EAAE,GAAG,SAAS,CAAC;CACtC;AAED,wBAAgB,qBAAqB,CAAC,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,GAAG,gBAAgB,CAoF5F;AAED,wBAAgB,yBAAyB,CACrC,UAAU,EAAE,MAAM,EAClB,aAAa,EAAE,MAAM,GAAG,SAAS,EACjC,SAAS,EAAE,MAAM,EAAE,GACpB,MAAM,EAAE,CAgGV;AAED,wBAAsB,iBAAiB,CAAC,cAAc,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,GAAG,MAAM,CAAC,CAMxF;AAED,wBAAsB,WAAW,CAAC,SAAS,CAAC,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE,YAAY,iBA4nC3E"}