omnius 1.0.863 → 1.0.864

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 (57) hide show
  1. package/assets/tray/omnius-checking.ico +0 -0
  2. package/assets/tray/omnius-checking.png +0 -0
  3. package/assets/tray/omnius-checking.svg +7 -4
  4. package/assets/tray/omnius-offline.ico +0 -0
  5. package/assets/tray/omnius-offline.png +0 -0
  6. package/assets/tray/omnius-offline.svg +6 -4
  7. package/assets/tray/omnius-online.ico +0 -0
  8. package/assets/tray/omnius-online.png +0 -0
  9. package/assets/tray/omnius-online.svg +5 -4
  10. package/dist/index.js +2 -2
  11. package/dist/library.js +11 -11
  12. package/dist/tui/native/worker.js +2283 -119
  13. package/docs/DISCOVERY.json +913 -75
  14. package/docs/DISCOVERY.md +23 -0
  15. package/docs/work-orders/desktop-projects.md +42 -0
  16. package/docs/work-orders/evidence/desktop-indicator-registration-20261016.json +23 -0
  17. package/docs/work-orders/evidence/inference-device-policy-readonly-20261016.json +56 -0
  18. package/docs/work-orders/evidence/instrument-100x30.png +0 -0
  19. package/docs/work-orders/evidence/instrument-120x36.png +0 -0
  20. package/docs/work-orders/evidence/instrument-160x48.png +0 -0
  21. package/docs/work-orders/evidence/instrument-60x20.png +0 -0
  22. package/docs/work-orders/evidence/instrument-80x24.png +0 -0
  23. package/docs/work-orders/evidence/instrument-clean-transcript-120x36.png +0 -0
  24. package/docs/work-orders/evidence/instrument-collapse-120x36.png +0 -0
  25. package/docs/work-orders/evidence/instrument-devices-expanded-120x36.png +0 -0
  26. package/docs/work-orders/evidence/instrument-divider-collapsed-160x48.png +0 -0
  27. package/docs/work-orders/evidence/instrument-divider-resized-160x48.png +0 -0
  28. package/docs/work-orders/evidence/instrument-help-scroll-120x36.png +0 -0
  29. package/docs/work-orders/evidence/instrument-popup-api-120x36.png +0 -0
  30. package/docs/work-orders/evidence/instrument-popup-api-input-120x36.png +0 -0
  31. package/docs/work-orders/evidence/instrument-popup-config-120x36.png +0 -0
  32. package/docs/work-orders/evidence/instrument-popup-dark-120x36.png +0 -0
  33. package/docs/work-orders/evidence/instrument-popup-dark-60x20.png +0 -0
  34. package/docs/work-orders/evidence/instrument-popup-light-120x36.png +0 -0
  35. package/docs/work-orders/evidence/instrument-popup-model-120x36.png +0 -0
  36. package/docs/work-orders/evidence/instrument-popup-network-120x36.png +0 -0
  37. package/docs/work-orders/evidence/instrument-production-terminal-live.json +7463 -1873
  38. package/docs/work-orders/evidence/instrument-report-broker-120x36.png +0 -0
  39. package/docs/work-orders/evidence/instrument-report-gpu-120x36.png +0 -0
  40. package/docs/work-orders/evidence/instrument-report-stats-120x36.png +0 -0
  41. package/docs/work-orders/evidence/instrument-semantic-components-20261016-updated.json +80 -0
  42. package/docs/work-orders/evidence/instrument-targeted-verification-20261016-updated.json +430 -0
  43. package/docs/work-orders/evidence/instrument-terminal-light-120x36.png +0 -0
  44. package/docs/work-orders/evidence/instrument-vim-composer-120x36.png +0 -0
  45. package/docs/work-orders/evidence/tray-native-host-checking.png +0 -0
  46. package/docs/work-orders/evidence/tray-native-host-dark.png +0 -0
  47. package/docs/work-orders/evidence/tray-native-host-offline.png +0 -0
  48. package/docs/work-orders/evidence/tray-native-host.png +0 -0
  49. package/docs/work-orders/evidence/tray-project-3.png +0 -0
  50. package/docs/work-orders/evidence/tray-project-4.png +0 -0
  51. package/docs/work-orders/evidence/tray-project-5.png +0 -0
  52. package/docs/work-orders/evidence/tray-projects-live.json +338 -0
  53. package/docs/work-orders/evidence/tray-projects-menu.png +0 -0
  54. package/docs/work-orders/instrument-command-surface-audit.md +215 -148
  55. package/docs/work-orders/instrument-tui.md +60 -12
  56. package/npm-shrinkwrap.json +2 -2
  57. package/package.json +1 -1
@@ -16,9 +16,12 @@ controls retain layout, pointer capture, wheel handling and keyboard interaction
16
16
  The same treatment applies to history, documents, instruments and selectors.
17
17
 
18
18
  Conversation remains the primary surface. At 110 columns and wider, an optional
19
- 38% instrument rail shows actual inference readings and a collapsed session
20
- section. On smaller terminals it starts closed; DETAILS opens a bounded stacked
21
- section. Explicit disclosure choices survive state updates and resizing. Unknown
19
+ instrument rail starts at 38% width, with actual inference readings and a
20
+ collapsed session section. Its one-cell divider drags to resize while preserving
21
+ at least 40 transcript columns. Dragging fully right collapses it to an arrow
22
+ handle; clicking or dragging that handle restores it. On smaller terminals it
23
+ starts closed; DETAILS opens a bounded stacked section. Width, collapse and
24
+ disclosure choices survive state updates and resizing. Unknown
22
25
  readings remain `--`; meters show their actual numeric reading. No invented
23
26
  telemetry, electrical diagrams or decorative targets are displayed.
24
27
 
@@ -30,9 +33,43 @@ No duplicate hint row or telemetry module is placed inside the editor.
30
33
  Configuration and help use native modal controls over the retained conversation.
31
34
  The underlying page is dimmed through text attributes, with its terminal colors,
32
35
  links and selection intact. Sparse hard-corner frames reserve their own border
33
- cells, with short connected horizontal and vertical arms. A brief opening effect
34
- extends those arms without moving content. It runs once on opening and has no
35
- idle animation loop.
36
+ cells, with short connected horizontal and vertical arms. Opening grows a small
37
+ center square to the dialog's bounds; closing reverses that motion. Content
38
+ layout stays fixed. Outside clicks dismiss the topmost dialog without activating
39
+ the underlying page. Animations stop after completion and on disposal.
40
+
41
+ Standalone session dashboards, broker reports, GPU policy, diagnostics and related
42
+ notices use their own native document modals. Read-only reports offer Refresh;
43
+ mutating commands never gain an implicit replay action. Scoped output capture
44
+ does not divert unrelated assistant/tool work or replace stdout. Stream lifecycle,
45
+ token/context counters and voice acknowledgements update structured chrome;
46
+ short notices expire after five seconds. Answers and real errors remain durable.
47
+
48
+ The sidebar's bottom device section occupies three rows until expanded. Its
49
+ native device list shows measured utilization and memory, with policy controls
50
+ using the existing `/gpu` selector and broker eligibility checks. Desktop NVIDIA
51
+ telemetry uses nvidia-smi; Jetson uses jtop and explicitly labels unified memory.
52
+ Missing sensors remain unknown. Detailed lease notices stay in STATUS's modal,
53
+ so they cannot displace device readings. These readings do not assert model-runner
54
+ placement. Hardware readings update on the existing collector; policy discovery
55
+ is coalesced and throttled separately.
56
+
57
+ JSON and fenced code share native syntax attributes over terminal-default colors.
58
+ JSON formatting changes whitespace only, preserving numeric and string lexemes.
59
+ Incomplete streams retain their source text. Headings and code disclosures remain
60
+ nested under their originating turn, with independent collapse state retained
61
+ during streamed updates. Bundled Tree-sitter grammars and a lexical fallback
62
+ require no grammar downloads.
63
+
64
+ `/vim [on|off|status]` controls the native composer, following the normal/insert
65
+ mode distinction in the [official Codex command documentation](https://learn.chatgpt.com/docs/developer-commands).
66
+ Esc enters normal mode; i/a/I/A return to insertion. Native buffer operations
67
+ support movement, counts, deletion/change/yank/paste, undo/redo, selection, find
68
+ and search without replacing the renderer or reparsing raw terminal input.
69
+ `/neovim [file]` and `/nvim` retain the project editor in an embedded native
70
+ terminal, with built-in file browsing, scoped terminal-default styling and no
71
+ automatic plugin downloads. Normal exit preserves unsaved buffers; only actual
72
+ application shutdown force-cleans the child process.
36
73
 
37
74
  Set `OMNIUS_TUI_REDUCE_MOTION=1` (or `REDUCE_MOTION=1`) to paint the final
38
75
  corner frame immediately.
@@ -56,8 +93,8 @@ corner frame immediately.
56
93
  coalesced NativeSession writes, including suspend/resume and shutdown.
57
94
 
58
95
  The [complete command and menu audit](instrument-command-surface-audit.md) maps
59
- all 118 catalog entries, their aliases and 354 signatures to shared presentation
60
- paths. It distinguishes 105 implemented commands from 13 planned entries and
96
+ all 119 catalog entries, their aliases and 356 signatures to shared presentation
97
+ paths. It distinguishes 106 implemented commands from 13 planned entries and
61
98
  indexes all 94 inspected selector call sites. Runtime menu items use the same
62
99
  components. The audit also repairs Help-to-editor handoff, the blocked network
63
100
  menu, nested pause ownership and remaining diagnostic/dashboard framing. It does
@@ -77,16 +114,24 @@ terminal screenshots and exercises keyboard and pointer interactions. It permits
77
114
  local model metadata and blocks inference and model-loading requests. It does
78
115
  not change the operator's terminal profile or GPU workloads.
79
116
 
80
- Final results: **599 tests across 75 focused files passed**, including all required
81
- daemon transport checks. The actual native renderer passed 16 component scenarios
82
- and four popup lifecycle scenarios. The production GNOME/VTE run passed all 12
83
- checks and produced 16 genuine screenshots, with default foreground/background
117
+ The earlier instrument baseline passed **599 tests across 75 focused files**,
118
+ including all required daemon transport checks. Current native checks also cover
119
+ the draggable sidebar, device monitor, code disclosures and Vim composer.
120
+ The updated curated regression set passed **767 CLI tests across 95 files plus
121
+ five Jetson telemetry tests**, with source hashes unchanged during execution.
122
+ The retained editor also passed a real installed Neovim 0.11.6 PTY check.
123
+ The production GNOME/VTE run passed all **16 checks with 23 actual screenshots**,
124
+ with default foreground/background
84
125
  intent on every visible cell in both dark and light profiles. No model generation
85
126
  or model loading was requested for this interface change.
86
127
 
87
128
  - [Targeted verification](evidence/instrument-targeted-verification-20261016.json)
129
+ - [Updated targeted verification](evidence/instrument-targeted-verification-20261016-updated.json)
88
130
  - [Native scenarios and timing limits](evidence/instrument-semantic-components-20261016.json)
89
131
  - [Actual terminal run](evidence/instrument-production-terminal-live.json)
132
+ - [Sidebar resizing](evidence/instrument-divider-resized-160x48.png), [collapsed handle](evidence/instrument-divider-collapsed-160x48.png)
133
+ - [Device monitor](evidence/instrument-devices-expanded-120x36.png), [GPU policy report](evidence/instrument-report-gpu-120x36.png)
134
+ - [Vim composer](evidence/instrument-vim-composer-120x36.png)
90
135
  - [Wide layout](evidence/instrument-120x36.png), [narrow layout](evidence/instrument-60x20.png)
91
136
  - [Dark popup](evidence/instrument-popup-dark-120x36.png), [light popup](evidence/instrument-popup-light-120x36.png)
92
137
  - [Configuration](evidence/instrument-popup-config-120x36.png), [network](evidence/instrument-popup-network-120x36.png), [nested input](evidence/instrument-popup-api-input-120x36.png)
@@ -106,3 +151,6 @@ The previous renderer and local inference repair are documented in
106
151
  [native-tui-renderer.md](native-tui-renderer.md). Its October 15 measurements are
107
152
  historical. This visual change does not claim a new model-inference benchmark or
108
153
  a pass of the entire CLI suite.
154
+
155
+ The [desktop indicator report](desktop-projects.md) covers the project dropdown,
156
+ monochrome assets and actual native menu-to-terminal launches.
@@ -1,12 +1,12 @@
1
1
  {
2
2
  "name": "omnius",
3
- "version": "1.0.863",
3
+ "version": "1.0.864",
4
4
  "lockfileVersion": 3,
5
5
  "requires": true,
6
6
  "packages": {
7
7
  "": {
8
8
  "name": "omnius",
9
- "version": "1.0.863",
9
+ "version": "1.0.864",
10
10
  "bundleDependencies": [
11
11
  "image-to-ascii"
12
12
  ],
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "omnius",
3
- "version": "1.0.863",
3
+ "version": "1.0.864",
4
4
  "description": "AI coding agent powered by open-source models (Ollama/vLLM) — interactive TUI with agentic tool-calling loop",
5
5
  "type": "module",
6
6
  "main": "./dist/library.js",