@8gi-foundation/8gent-code 0.6.0

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.
package/README.md ADDED
@@ -0,0 +1,714 @@
1
+ <p align="center">
2
+ <img src=".github/assets/readme-header.png" width="100%" alt="8gent Code" />
3
+ </p>
4
+
5
+ <p align="center">
6
+ <strong>The kernel of the <a href="https://8gent.world">8gent ecosystem</a>.</strong><br />
7
+ Open source autonomous coding agent powered by local LLMs or free cloud models.<br />
8
+ No API keys. No usage caps. No cloud dependency.
9
+ </p>
10
+
11
+ <br />
12
+
13
+ <p align="center">
14
+ <a href="https://8gentjr.com"><img src="https://img.shields.io/badge/Jr-Live-2D8A56?style=for-the-badge&labelColor=1A1612" alt="Jr Live" /></a>
15
+ <a href="https://8gentos.com"><img src="https://img.shields.io/badge/OS-In_Dev-E8610A?style=for-the-badge&labelColor=1A1612" alt="OS In Dev" /></a>
16
+ <a href="https://github.com/8gi-foundation/8gent-code"><img src="https://img.shields.io/badge/Code-Open_Source-2D8A56?style=for-the-badge&labelColor=1A1612" alt="Code Open Source" /></a>
17
+ <a href="https://8gent.world"><img src="https://img.shields.io/badge/World-Live-2D8A56?style=for-the-badge&labelColor=1A1612" alt="World Live" /></a>
18
+ <a href="https://8gent.games"><img src="https://img.shields.io/badge/Games-Live-2D8A56?style=for-the-badge&labelColor=1A1612" alt="Games Live" /></a>
19
+ </p>
20
+
21
+ <p align="center">
22
+ <a href="https://www.apache.org/licenses/LICENSE-2.0"><img src="https://img.shields.io/badge/License-Apache_2.0-E8610A?style=for-the-badge&labelColor=1A1612" alt="Apache 2.0 License" /></a>
23
+ <a href="https://8gent.dev"><img src="https://img.shields.io/badge/version-0.3.0-2D8A56?style=for-the-badge&labelColor=1A1612" alt="v2.0.0" /></a>
24
+ <a href="https://eight-vessel.fly.dev"><img src="https://img.shields.io/badge/daemon-Fly.io_Amsterdam-E8610A?style=for-the-badge&labelColor=1A1612" alt="Daemon" /></a>
25
+ </p>
26
+
27
+ <br />
28
+
29
+ ---
30
+
31
+ <br />
32
+
33
+ ## The Ecosystem
34
+
35
+ <p align="center"><sub>6 products &nbsp;·&nbsp; 6 domains &nbsp;·&nbsp; 1 constitution</sub></p>
36
+
37
+ <br />
38
+
39
+ <table>
40
+ <tr>
41
+ <td valign="top" width="33%">
42
+
43
+ **8gent OS** -[8gentos.com](https://8gentos.com)<br />
44
+ <sub>Parent site. Paid product. Revenue engine.</sub>
45
+
46
+ **8gent Code** -[8gent.dev](https://8gent.dev)<br />
47
+ <sub>Open source developer agent. Free on-ramp. <em>(this repo)</em></sub>
48
+
49
+ </td>
50
+ <td valign="top" width="33%">
51
+
52
+ **8gent** -[8gent.app](https://8gent.app)<br />
53
+ <sub>Consumer GUI client for the OS.</sub>
54
+
55
+ **8gent World** -[8gent.world](https://8gent.world)<br />
56
+ <sub>Ecosystem story, docs, <a href="https://8gent.world/media/decks">14 presentation decks</a>.</sub>
57
+
58
+ </td>
59
+ <td valign="top" width="33%">
60
+
61
+ **8gent Games** -[8gent.games](https://8gent.games)<br />
62
+ <sub>Agent simulation playground.</sub>
63
+
64
+ **8gent Jr** -[8gentjr.com](https://8gentjr.com)<br />
65
+ <sub>AI assistant for kids. Accessibility first. Free forever.</sub>
66
+
67
+ </td>
68
+ </tr>
69
+ </table>
70
+
71
+ <p align="center">
72
+ <sub><a href="https://8gent.world/constitution">Constitution</a> &nbsp;·&nbsp; <a href="https://8gent.world/inspirations">Inspirations</a></sub>
73
+ </p>
74
+
75
+ <br />
76
+
77
+ ---
78
+
79
+ <br />
80
+
81
+ ## 8GI Foundation
82
+
83
+ 8gent Code is the technology layer of the **8GI Foundation** - the autonomous collective intelligence that governs the 8gent ecosystem. 8GI is not a company. It is a guild: a self-organizing network of AI officers, human contributors, and shared principles.
84
+
85
+ Engineers who contribute to 8gent Code learn agentic organization patterns firsthand - how autonomous agents coordinate, govern themselves, and scale without traditional management hierarchies. See the [Guild Deck](https://8gent.world/media/decks) for the full vision.
86
+
87
+ All governance docs, decks, and the constitution live at [8gent.world](https://8gent.world). Source-of-truth markdown for governance (including security and onboarding) is in the [`8gi-governance`](https://github.com/8gi-foundation/8gi-governance) repository; static deck assets ship with [`8gent-world`](https://github.com/8gi-foundation/8gent-world) under `public/media/`.
88
+
89
+ ### The Board - 8 Seats of the Inner Circle
90
+
91
+ The 8GI board consists of AI officers, not humans. James Spalding serves as Founder and Visionary.
92
+
93
+ | Seat | Officer | Role |
94
+ |:-----|:--------|:-----|
95
+ | **8EO** | AI James | Eight Executive Officer - strategy, coordination, ecosystem oversight |
96
+ | **8TO** | Rishi | Eight Technology Officer - architecture, infrastructure, technical direction |
97
+ | **8PO** | Samantha | Eight Product Officer - product vision, UX, user advocacy |
98
+ | **8DO** | Moira | Eight Design Officer - brand, visual identity, design systems |
99
+ | **8SO** | Karen | Eight Security Officer - policy, compliance, threat modeling |
100
+ | **8MO** | *Pending* | Eight Marketing Officer |
101
+ | **8CO** | *Pending* | Eight Community Officer |
102
+ | **8GO** | *Pending* | Eight Governance Officer |
103
+
104
+ ### The Lotus Model
105
+
106
+ 8GI scales through the Lotus structure: **1-8-64-512**. One founder. Eight AI officers (the inner circle). 64 working vessels (specialized agents). 512 edge nodes (community contributors and autonomous tasks). Each layer multiplies capacity without multiplying complexity.
107
+
108
+ ### Constitution
109
+
110
+ The [10 Articles of the 8gent Constitution](https://8gent.world/constitution) govern all decisions across every product and every agent in the ecosystem.
111
+
112
+ ### Control Plane Architecture
113
+
114
+ The autonomous vessel infrastructure lives in two packages:
115
+
116
+ - `packages/board-plane/` - the control plane that coordinates board-level decisions and vessel orchestration
117
+ - `packages/board-vessel/` - the blueprint pattern for spawning autonomous AI officer vessels
118
+
119
+ These implement the board's ability to operate as a persistent, self-governing collective.
120
+
121
+ ### GitHub and Community
122
+
123
+ All code lives under [github.com/8gi-foundation](https://github.com/8gi-foundation).
124
+
125
+ | Repo | Role |
126
+ |------|------|
127
+ | **[8gent](https://github.com/8gi-foundation/8gent)** (8gent.app) | The front door. Dashboard. Auth. Billing. User management. |
128
+ | **[8gi-control-plane](https://github.com/8gi-foundation/8gi-control-plane)** | The brain. Model routing. Rate limiting. Token tracking. |
129
+ | **[8gent-vessel](https://github.com/8gi-foundation/8gent-vessel)** | The body. Compute. Sandboxes. Storage. Health. |
130
+ | **[8gent-code](https://github.com/8gi-foundation/8gent-code)** | The kernel. What runs inside every vessel. *(this repo)* |
131
+ | **[8gent-OS](https://github.com/8gi-foundation/8gent-OS)** | The personal layer on top of the kernel. |
132
+ | **[8gi-governance](https://github.com/8gi-foundation/8gi-governance)** | The constitution. Board decisions. Member registry. |
133
+ | **[8gent-world](https://github.com/8gi-foundation/8gent-world)** | Ecosystem story, docs, media. |
134
+ | **[8gent-dev](https://github.com/8gi-foundation/8gent-dev)** | Developer portal. |
135
+ | **[8gent-games](https://github.com/8gi-foundation/8gent-games)** | Agent simulation playground. |
136
+ | **[8gent-telegram-app](https://github.com/8gi-foundation/8gent-telegram-app)** | Jr Telegram interface. |
137
+
138
+ The 8GI Foundation Discord server is the primary community hub for contributors and guild members.
139
+
140
+ <br />
141
+
142
+ ---
143
+
144
+ <br />
145
+
146
+ ## Quick Start
147
+
148
+ ```bash
149
+ npm install -g @8gi-foundation/8gent-code
150
+ 8gent
151
+ ```
152
+
153
+ That's it. Ollama runs locally by default - if you don't have it, 8gent will guide you through setup on first launch.
154
+
155
+ ## Quick Start (Harness Smoke)
156
+
157
+ ```bash
158
+ # 1. Clone and install
159
+ git clone https://github.com/8gi-foundation/8gent-code.git
160
+ cd 8gent-code
161
+ bun install
162
+
163
+ # 2. Verify CLI works
164
+ bun run cli -- --version
165
+ # Expected: 0.x.x (8gent Code)
166
+
167
+ # 3. Verify harness typecheck
168
+ bun run check:harness
169
+ # Expected: exit 0 (or type errors listed — not a crash)
170
+ ```
171
+
172
+ If Ollama is not installed, 8gent will prompt you on first run. For offline/CI use:
173
+ ```bash
174
+ INFERENCE_MODE=proxy bun run cli -- --version
175
+ ```
176
+
177
+ <br />
178
+
179
+ ---
180
+
181
+ <br />
182
+
183
+ ## Why 8gent exists
184
+
185
+ Token vendors control access to intelligence through pricing tiers, rate limits, and API keys.
186
+ That is a business model, not a law of nature. It is also not the only option.
187
+
188
+ 8gent runs locally, privately, and for free. No credit card. No usage cap. No cloud dependency required.
189
+
190
+ Every policy that governs what the agent can do is a YAML file you can read, edit, and override.
191
+ Every memory the agent stores is a SQLite database on your own disk. Nothing phones home.
192
+
193
+ Self-improvement: the autoresearch loop runs benchmarks, mutates the system prompt, and promotes what works. This runs locally.
194
+ Your agent runs locally. Your data never leaves your machine. Every policy is readable YAML.
195
+ No central vendor captures that value.
196
+
197
+ The floor is zero cost. The ceiling is what a self-improving local agent can learn from your codebase.
198
+
199
+ Try it: `npm install -g @8gi-foundation/8gent-code && 8gent`
200
+
201
+ ### From source (contributors)
202
+
203
+ ```bash
204
+ git clone https://github.com/8gi-foundation/8gent-code.git && cd 8gent-code && bun install
205
+ bun run tui
206
+ ```
207
+
208
+ <br />
209
+
210
+ ---
211
+
212
+ <br />
213
+
214
+ ## What Makes This Different
215
+
216
+ <table>
217
+ <tr>
218
+ <td valign="top" width="50%">
219
+
220
+ **Local-first, free by default**<br />
221
+ <sub>Runs entirely on your machine. Cloud models (OpenRouter free tier) are opt-in. No telemetry, no API keys to start.</sub>
222
+
223
+ <br />
224
+
225
+ **Model-agnostic**<br />
226
+ <sub>Auto-selects from best free models on OpenRouter. Runs Qwen 3.5 via Ollama locally. Task router classifies prompts (code / reasoning / simple / creative) and picks the best model automatically.</sub>
227
+
228
+ <br />
229
+
230
+ **Eight kernel**<br />
231
+ <sub>Persistent daemon deployed on Fly.io Amsterdam (<a href="https://eight-vessel.fly.dev">eight-vessel.fly.dev</a>). WebSocket protocol, 4-strategy retry loop, session persistence across reconnections.</sub>
232
+
233
+ <br />
234
+
235
+ **NemoClaw policy engine**<br />
236
+ <sub>YAML-based, deny-by-default, rebuilt from scratch. 11 default rules with approval gates for secrets, destructive ops, network, git, and file access. Headless and infinite modes for autonomous operation.</sub>
237
+
238
+ </td>
239
+ <td valign="top" width="50%">
240
+
241
+ **8 Powers**<br />
242
+ <sub>Memory, parallel worktrees, NemoClaw policy, self-evolution, self-healing, entrepreneurship, AST blast radius, and browser access. Not plugins. Built-in.</sub>
243
+
244
+ <br />
245
+
246
+ **HyperAgent meta-improvement**<br />
247
+ <sub>Metacognitive self-modification. The agent can improve how it improves -meta-config is editable while the evaluation protocol stays human-controlled.</sub>
248
+
249
+ <br />
250
+
251
+ **AutoResearch**<br />
252
+ <sub>Overnight improvement loops (Karpathy-style). Runs benchmarks, mutates system prompts, re-tests. Meta-optimizer also tunes few-shots, model routing, and grading weights.</sub>
253
+
254
+ <br />
255
+
256
+ **Voice**<br />
257
+ <sub>8 neural AI voices (KittenTTS, free, local) + full-duplex via Moshi on Apple Silicon. Auto-installed during onboarding. <code>/voice chat</code> to start.</sub>
258
+
259
+ <br />
260
+
261
+ **AST-first navigation** &nbsp;·&nbsp; **Multi-agent orchestration** &nbsp;·&nbsp; **Telegram portal**
262
+
263
+ </td>
264
+ </tr>
265
+ </table>
266
+
267
+ <br />
268
+
269
+ ---
270
+
271
+ <br />
272
+
273
+ ## The 8 Powers
274
+
275
+ <table>
276
+ <tr>
277
+ <td valign="top" width="25%">
278
+
279
+ **Memory**<br />
280
+ <sub><code>packages/memory/</code></sub><br />
281
+ <sub>Dual-layer episodic + semantic memory, SQLite + FTS5, Ollama embeddings, procedural memory, health monitoring, contradiction detection, consolidation, lease-based job queue</sub>
282
+
283
+ </td>
284
+ <td valign="top" width="25%">
285
+
286
+ **Worktree**<br />
287
+ <sub><code>packages/orchestration/</code></sub><br />
288
+ <sub>Multi-agent parallel execution via git worktrees, max 4 concurrent, filesystem messaging, macro-actions, delegation</sub>
289
+
290
+ </td>
291
+ <td valign="top" width="25%">
292
+
293
+ **Policy**<br />
294
+ <sub><code>packages/permissions/</code></sub><br />
295
+ <sub>NemoClaw YAML policy engine, 11 default rules, approval gates, headless mode, infinite mode, dangerous command detection</sub>
296
+
297
+ </td>
298
+ <td valign="top" width="25%">
299
+
300
+ **Evolution**<br />
301
+ <sub><code>packages/self-autonomy/</code></sub><br />
302
+ <sub>Post-session reflection, Bayesian skill confidence, HyperAgent meta-mutation, skill compounding (tasks become reusable skills), KittenTTS voice onboarding</sub>
303
+
304
+ </td>
305
+ </tr>
306
+ <tr>
307
+ <td valign="top" width="25%">
308
+
309
+ **Healing**<br />
310
+ <sub><code>packages/validation/</code></sub><br />
311
+ <sub>Checkpoint-verify-revert loop, git-stash atomic snapshots, failure log</sub>
312
+
313
+ </td>
314
+ <td valign="top" width="25%">
315
+
316
+ **Entrepreneurship**<br />
317
+ <sub><code>packages/proactive/</code></sub><br />
318
+ <sub>GitHub bounty/help-wanted scanner, capability matcher, opportunity pipeline</sub>
319
+
320
+ </td>
321
+ <td valign="top" width="25%">
322
+
323
+ **AST**<br />
324
+ <sub><code>packages/ast-index/</code></sub><br />
325
+ <sub>Blast radius engine, import dependency graph, test file mapping, change impact estimation</sub>
326
+
327
+ </td>
328
+ <td valign="top" width="25%">
329
+
330
+ **Browser**<br />
331
+ <sub><code>packages/tools/browser/</code></sub><br />
332
+ <sub>Lightweight web access via fetch + DuckDuckGo HTML scraping, disk cache, no headless deps</sub>
333
+
334
+ </td>
335
+ </tr>
336
+ </table>
337
+
338
+ <br />
339
+
340
+ ---
341
+
342
+ <br />
343
+
344
+ ## Companion System
345
+
346
+ Every session spawns a unique companion. Your coding history becomes a collectible deck.
347
+
348
+ - **40 species** across 5 rarity tiers (Common 60% to Legendary 1%)
349
+ - **10 elements** inspired by MTG color pie (Void, Ember, Aether, Verdant, Radiant, Chrome, Prism, Frost, Thunder, Shadow)
350
+ - **29 accessories** from Pokeball to Triforce to One Ring
351
+ - **6 stats** per companion (DEBUG, CHAOS, WISDOM, PATIENCE, SNARK, ARCANA)
352
+ - **1% shiny** chance
353
+ - **Collection deck** persists at `~/.8gent/companion-deck.json`
354
+ - **macOS dock pet** spawns with companion's name and colors
355
+
356
+ ```bash
357
+ /pet start # Spawn companion on dock
358
+ /pet deck # View your collection
359
+ /pet card # Roll a new card
360
+ ```
361
+
362
+ See [packages/pet/README.md](packages/pet/README.md) for the full bestiary.
363
+
364
+ <br />
365
+
366
+ ---
367
+
368
+ <br />
369
+
370
+ ## Presentations
371
+
372
+ | Deck | Link |
373
+ |------|------|
374
+ | **npm Launch** | [8gent.world/media/decks/npm-launch](https://8gent.world/media/decks/npm-launch) |
375
+ | **Lil Eight Pets** | [8gent.world/media/decks/lil-eight](https://8gent.world/media/decks/lil-eight) |
376
+ | **Companion System** | [8gent.world/media/decks/companion-system](https://8gent.world/media/decks/companion-system) |
377
+ | **Code Roadmap** | [8gent.world/media/decks/code-roadmap](https://8gent.world/media/decks/code-roadmap) |
378
+ | **All Decks** | [8gent.world/media/decks](https://8gent.world/media/decks) |
379
+
380
+ <br />
381
+
382
+ ---
383
+
384
+ <br />
385
+
386
+ ## Voice
387
+
388
+ Two modes: half-duplex voice chat and full-duplex conversation.
389
+
390
+ **Half-duplex** (`/voice chat`) - listen, transcribe, think, speak, repeat. Requires sox and whisper.cpp:
391
+
392
+ ```bash
393
+ brew install sox whisper-cpp
394
+ ```
395
+
396
+ **Neural TTS** - 8 local AI voices via KittenTTS (free, no API key). Auto-offered during onboarding:
397
+
398
+ | Voice | Gender | Character |
399
+ |:------|:-------|:----------|
400
+ | **Bruno** (default) | Male | Warm, authoritative |
401
+ | Bella | Female | Warm, clear |
402
+ | Jasper | Male | Crisp, technical |
403
+ | Luna | Female | Soft, creative |
404
+ | Rosie | Female | Bright, energetic |
405
+ | Hugo | Male | Neutral, steady |
406
+ | Kiki | Female | Light, friendly |
407
+ | Leo | Male | Rich, expressive |
408
+
409
+ Falls back to macOS `say` (Moira, Daniel, Samantha, Karen, Rishi) if KittenTTS is not installed.
410
+
411
+ **Full-duplex** - simultaneous listen and speak via Moshi (Kyutai) on Apple Silicon. Requires `pip install moshi`. Backend auto-detected.
412
+
413
+ <sub>Status bar shows: <strong>VOICE CHAT (listening)</strong> / <strong>SPEAKING</strong> / <strong>THINKING</strong></sub>
414
+
415
+ <br />
416
+
417
+ ---
418
+
419
+ <br />
420
+
421
+ ## How It Works
422
+
423
+ ```
424
+ User prompt
425
+ -> BMAD planner (structured task decomposition)
426
+ -> Multi-agent orchestration (sub-agents in worktrees)
427
+ -> Toolshed (MCP, LSP, shell, AST, filesystem)
428
+ -> Execution + validation (self-healing loop)
429
+ -> Result
430
+ ```
431
+
432
+ <sub>The agent decomposes work, delegates to sub-agents, validates output against test suites, and reports back. It uses the BMAD method for planning and AST-level symbol retrieval to keep token usage minimal.</sub>
433
+
434
+ <br />
435
+
436
+ ---
437
+
438
+ <br />
439
+
440
+ ## Benchmarks
441
+
442
+ <p align="center"><sub>Execution-graded tests across professional domains. All local inference via Ollama.<br />Code compiles and runs against <code>bun:test</code> suites, or it fails. No string matching, no vibes.</sub></p>
443
+
444
+ <br />
445
+
446
+ | ID | Domain | Task | Score |
447
+ |:---|:-------|:-----|------:|
448
+ | BT001 | Software Engineering | SaaS Auth: JWT, Roles, Rate Limiting | **94** |
449
+ | BT002 | Software Engineering | Event-Driven Architecture: Pub/Sub, DLQ, Retry | **92** |
450
+ | BT003 | Data Engineering | Stream Processing Pipeline | **100** |
451
+ | BT005 | Software Engineering | Typed State Machine: Guards, Actions | **92** |
452
+ | BT007 | Digital Marketing | SEO Audit Engine: Scoring, Core Web Vitals | **96** |
453
+ | BT011 | Video Production | Scene Graph, Timeline, FFmpeg CLI | **100** |
454
+ | BT012 | Music Technology | Notes, Chords, Scales, Progressions | **81** |
455
+ | BT014 | AI Consulting | Assessment Report Generator | **95** |
456
+
457
+ <sub>Additional categories: long-horizon (LH001–LH005), agentic (TC001–MR001), fullstack (FS001–FS003), UI design (UI001–UI008), ability showcase.</sub>
458
+
459
+ ```bash
460
+ bun run benchmark:v2 # single pass
461
+ CATEGORY=battle-test bun run benchmark:loop # autoresearch loop
462
+ ```
463
+
464
+ <sub>Full results: <a href="benchmarks/README.md">benchmarks/README.md</a> &nbsp;·&nbsp; Model shootout: <a href="docs/MODEL-SHOOTOUT.md">docs/MODEL-SHOOTOUT.md</a></sub>
465
+
466
+ <br />
467
+
468
+ ---
469
+
470
+ <br />
471
+
472
+ ## Project Structure
473
+
474
+ <table>
475
+ <tr>
476
+ <td valign="top" width="50%">
477
+
478
+ ### Apps
479
+
480
+ ```
481
+ apps/
482
+ tui/ Ink v6 terminal UI (main interface)
483
+ clui/ Tauri 2.0 desktop overlay (scaffolded)
484
+ dashboard/ Next.js admin panel
485
+ debugger/ Session debugger
486
+ demos/ Remotion video generation
487
+ installer/ Interactive install wizard
488
+ ```
489
+
490
+ </td>
491
+ <td valign="top" width="50%">
492
+
493
+ ### Packages
494
+
495
+ ```
496
+ packages/
497
+ eight/ Core agent engine (Vercel AI SDK)
498
+ daemon/ Persistent vessel daemon (Fly.io)
499
+ ai/ Provider abstraction (Ollama, OpenRouter)
500
+ memory/ SQLite + FTS5 persistent memory
501
+ orchestration/ WorktreePool, macro actions
502
+ permissions/ NemoClaw YAML policy engine
503
+ self-autonomy/ Evolution, reflection, HyperAgent
504
+ validation/ Self-healing executor
505
+ proactive/ Business agents, opportunity scanner
506
+ ast-index/ Blast radius engine
507
+ tools/ Browser, actuators, filesystem, shell
508
+ voice/ STT (whisper.cpp) + TTS (KittenTTS/macOS) + full-duplex (Moshi)
509
+ kernel/ RL fine-tuning pipeline (off by default)
510
+ personality/ Brand voice, "Infinite Gentleman"
511
+ telegram/ Telegram bot portal
512
+ auth/ Clerk auth + GitHub integration
513
+ db/ Convex reactive database
514
+ control-plane/ Multi-tenant management
515
+ board-plane/ Board-level vessel orchestration
516
+ board-vessel/ Autonomous AI officer blueprint
517
+ ```
518
+
519
+ </td>
520
+ </tr>
521
+ </table>
522
+
523
+ <sub>Additional directories: <code>benchmarks/</code> execution-graded benchmarks + autoresearch &nbsp;·&nbsp; <code>bin/</code> CLI entry points &nbsp;·&nbsp; <code>docs/</code> architecture docs</sub>
524
+
525
+ <br />
526
+
527
+ ---
528
+
529
+ <br />
530
+
531
+ ## Roadmap
532
+
533
+ <table>
534
+ <tr>
535
+ <td valign="top" width="33%">
536
+
537
+ ### Now
538
+
539
+ - **Harness hardening** - brain/hands isolation, context compression, and sub-agent spawn protocol landed. Security audit on spawn governance in progress (#1409).
540
+ - **Skill compounding** - tasks become reusable skills. Refining quality gate and promotion logic.
541
+ - **Voice mode** - KittenTTS neural voices integrated. Full-duplex Moshi backend working on Apple Silicon. Evaluating PersonaPlex (NVIDIA) for persona-controlled voice.
542
+ - **AST-index expansion** - symbol-level call edges, reachability queries, dead-code detection (#1392-#1394).
543
+
544
+ </td>
545
+ <td valign="top" width="33%">
546
+
547
+ ### Next
548
+
549
+ - **MCP server support** - expose 8gent tools as MCP servers for external agent consumption
550
+ - **Context window compaction** - smarter compression beyond token threshold
551
+ - **Extension system** - ExtensionCrafter for autonomous source-to-extension generation (#1254)
552
+ - [HyperAgent meta-improvement loop](docs/HYPERAGENT-SPEC.md)
553
+ - [Kernel fine-tuning pipeline](docs/KERNEL-FINETUNING.md) activation
554
+
555
+ </td>
556
+ <td valign="top" width="33%">
557
+
558
+ ### Later
559
+
560
+ - Desktop client (Tauri 2.0, scaffolded in `apps/clui/`)
561
+ - Multi-tenant control plane via 8GI gateway
562
+ - Full autonomous issue resolution
563
+ - Personal LoRA from session training pairs
564
+ - 8gent Jr voice vessel (KittenTTS server-side for mobile clients)
565
+
566
+ </td>
567
+ </tr>
568
+ </table>
569
+
570
+ <br />
571
+
572
+ ---
573
+
574
+ <br />
575
+
576
+ ## Slash Commands
577
+
578
+ <table>
579
+ <tr>
580
+ <td valign="top" width="50%">
581
+
582
+ | Command | What it does |
583
+ |:--------|:-------------|
584
+ | `/voice chat` | Start voice conversation mode |
585
+ | `/voice start` | Push-to-talk recording |
586
+ | `/model <name>` | Switch LLM model |
587
+ | `/board` | Kanban task board |
588
+ | `/predict` | Confidence-scored step predictions |
589
+ | `/momentum` | Velocity stats |
590
+ | `/evidence` | Session evidence summary |
591
+
592
+ </td>
593
+ <td valign="top" width="50%">
594
+
595
+ | Command | What it does |
596
+ |:--------|:-------------|
597
+ | `/history` | Browse past sessions |
598
+ | `/resume` | Resume a previous session |
599
+ | `/compact` | Compact current session |
600
+ | `/github` | GitHub integration |
601
+ | `/auth status` | Check auth state |
602
+ | `/debug` | Session inspector |
603
+ | `/deploy <target>` | Deploy to Vercel/Railway/Fly |
604
+ | `/throughput` | Token throughput stats |
605
+ | `/scorecard` | Ability scorecard metrics |
606
+ | `/soul` | Current persona calibration |
607
+ | `/router` | Task router + model selection |
608
+ | `/music` | Toggle lofi music (ADHD mode) |
609
+ | `/rename` | Rename the current session |
610
+
611
+ </td>
612
+ </tr>
613
+ </table>
614
+
615
+ <br />
616
+
617
+ ---
618
+
619
+ <br />
620
+
621
+ ## Documentation
622
+
623
+ <details>
624
+ <summary><strong>Architecture &amp; Specs</strong></summary>
625
+
626
+ <br />
627
+
628
+ | Doc | What it covers |
629
+ |:----|:---------------|
630
+ | [SOUL.md](SOUL.md) | Agent persona and principles |
631
+ | [CLAUDE.md](CLAUDE.md) | Dev conventions, design system, repo rules |
632
+ | [docs/HYPERAGENT-SPEC.md](docs/HYPERAGENT-SPEC.md) | HyperAgent metacognitive self-modification spec |
633
+ | [docs/MODEL-SHOOTOUT.md](docs/MODEL-SHOOTOUT.md) | Local vs cloud model comparison results |
634
+ | [docs/MEMORY-SPEC.md](docs/MEMORY-SPEC.md) | Memory layer architecture and API reference |
635
+ | [docs/KERNEL-FINETUNING.md](docs/KERNEL-FINETUNING.md) | RL fine-tuning pipeline |
636
+ | [docs/PERSONALIZATION.md](docs/PERSONALIZATION.md) | 5-layer personalization system |
637
+ | [docs/TOOLSHED.md](docs/TOOLSHED.md) | Capability discovery and skill registry |
638
+ | [docs/permissions.md](docs/permissions.md) | Policy engine and approval gates |
639
+ | [docs/BRANCH-DECISIONS.md](docs/BRANCH-DECISIONS.md) | Architecture decision log |
640
+ | [CONTRIBUTING.md](CONTRIBUTING.md) | How to contribute |
641
+
642
+ </details>
643
+
644
+ <details>
645
+ <summary><strong>External Resources</strong></summary>
646
+
647
+ <br />
648
+
649
+ | Resource | Link |
650
+ |:---------|:-----|
651
+ | 8gent Constitution | [8gent.world/constitution](https://8gent.world/constitution) |
652
+ | Presentation Decks | [8gent.world/media/decks](https://8gent.world/media/decks) |
653
+ | Architecture Inspirations | [8gent.world/inspirations](https://8gent.world/inspirations) |
654
+
655
+ </details>
656
+
657
+ <br />
658
+
659
+ ---
660
+
661
+ <br />
662
+
663
+ ## Inspirations
664
+
665
+ Architecture credits. These projects informed specific parts of 8gent's design.
666
+
667
+ <table>
668
+ <tr>
669
+ <td valign="top" width="50%">
670
+
671
+ - [Hermes by ArcadeAI](https://github.com/ArcadeAI/hermes) -persistent memory and self-evolution patterns
672
+ - [CashClaw](https://github.com/nicepkg/CashClaw) -autonomous work discovery and value generation
673
+ - NemoClaw -policy-driven governance and approval gate architecture
674
+ - HyperAgents (Meta FAIR, March 2026) -metacognitive self-modification
675
+ - Hypothesis Loop -atomic commit-verify-revert development cycle
676
+
677
+ </td>
678
+ <td valign="top" width="50%">
679
+
680
+ - Blast Radius Engine -AST-based change impact estimation
681
+ - Claude Code -worktree isolation pattern for parallel agent execution
682
+ - Karpathy's autoresearch methodology -iterative prompt mutation and meta-optimization
683
+ - [SoulSpec](https://github.com/OpenSoul-org/SoulSpec) -agent persona standard
684
+ - [usecomputer](https://github.com/remorses/usecomputer) -cross-platform desktop automation via native Zig N-API
685
+ - [Quitty](https://github.com/iad1tya/Quitty) -process management and resource conservation UX
686
+
687
+ </td>
688
+ </tr>
689
+ </table>
690
+
691
+ <sub>Full list at <a href="https://8gent.world/inspirations">8gent.world/inspirations</a></sub>
692
+
693
+ <br />
694
+
695
+ ---
696
+
697
+ <br />
698
+
699
+ <p align="center">
700
+ <strong>Apache 2.0</strong> - James Spalding, Founder and Visionary
701
+ </p>
702
+
703
+ <p align="center">
704
+ <a href="https://x.com/8gentapp">X / Twitter</a> &nbsp;·&nbsp;
705
+ <a href="https://github.com/8gi-foundation/8gent-code">GitHub</a> &nbsp;·&nbsp;
706
+ <a href="https://8gent.dev">8gent.dev</a> &nbsp;·&nbsp;
707
+ <a href="https://8gent.world">8gent.world</a>
708
+ </p>
709
+
710
+ <br />
711
+
712
+ <p align="center">
713
+ <sub>Your OS. Your rules. Your AI.</sub>
714
+ </p>