@theihtisham/agent-shadow-brain 5.1.1 → 5.2.0
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 +1159 -1095
- package/dist/adapters/copilot.d.ts +14 -0
- package/dist/adapters/copilot.d.ts.map +1 -0
- package/dist/adapters/copilot.js +93 -0
- package/dist/adapters/copilot.js.map +1 -0
- package/dist/adapters/cursor.d.ts +13 -0
- package/dist/adapters/cursor.d.ts.map +1 -0
- package/dist/adapters/cursor.js +96 -0
- package/dist/adapters/cursor.js.map +1 -0
- package/dist/adapters/index.d.ts +4 -1
- package/dist/adapters/index.d.ts.map +1 -1
- package/dist/adapters/index.js +7 -3
- package/dist/adapters/index.js.map +1 -1
- package/dist/adapters/windsurf.d.ts +14 -0
- package/dist/adapters/windsurf.d.ts.map +1 -0
- package/dist/adapters/windsurf.js +107 -0
- package/dist/adapters/windsurf.js.map +1 -0
- package/dist/brain/global-brain.d.ts +56 -0
- package/dist/brain/global-brain.d.ts.map +1 -0
- package/dist/brain/global-brain.js +324 -0
- package/dist/brain/global-brain.js.map +1 -0
- package/dist/brain/l0-cache.d.ts +40 -0
- package/dist/brain/l0-cache.d.ts.map +1 -0
- package/dist/brain/l0-cache.js +176 -0
- package/dist/brain/l0-cache.js.map +1 -0
- package/dist/brain/mcp-server.d.ts +4 -0
- package/dist/brain/mcp-server.d.ts.map +1 -1
- package/dist/brain/mcp-server.js +91 -0
- package/dist/brain/mcp-server.js.map +1 -1
- package/dist/brain/session-hooks.d.ts +28 -0
- package/dist/brain/session-hooks.d.ts.map +1 -0
- package/dist/brain/session-hooks.js +475 -0
- package/dist/brain/session-hooks.js.map +1 -0
- package/dist/brain/subconscious.d.ts +34 -0
- package/dist/brain/subconscious.d.ts.map +1 -0
- package/dist/brain/subconscious.js +275 -0
- package/dist/brain/subconscious.js.map +1 -0
- package/dist/cli.js +167 -2
- package/dist/cli.js.map +1 -1
- package/dist/types.d.ts +128 -1
- package/dist/types.d.ts.map +1 -1
- package/package.json +46 -10
package/README.md
CHANGED
|
@@ -1,1095 +1,1159 @@
|
|
|
1
|
-
<div align="center">
|
|
2
|
-
|
|
3
|
-
<img src="banner.svg" alt="Agent Shadow Brain v5.
|
|
4
|
-
|
|
5
|
-
<br/>
|
|
6
|
-
|
|
7
|
-
[](https://www.npmjs.com/package/@theihtisham/agent-shadow-brain)
|
|
8
|
-
[](https://www.npmjs.com/package/@theihtisham/agent-shadow-brain)
|
|
9
|
-
[](https://www.npmjs.com/package/@theihtisham/agent-shadow-brain)
|
|
8
|
+
[](https://www.npmjs.com/package/@theihtisham/agent-shadow-brain)
|
|
9
|
+
[](https://www.npmjs.com/package/@theihtisham/agent-shadow-brain)
|
|
10
|
+
[](https://bundlephobia.com/package/@theihtisham/agent-shadow-brain)
|
|
11
|
+
[](https://github.com/theihtisham/agent-shadow-brain/blob/master/LICENSE)
|
|
12
|
+
[](https://nodejs.org/)
|
|
13
|
+
[](https://www.typescriptlang.org/)
|
|
14
|
+
[](https://github.com/theihtisham/agent-shadow-brain/stargazers)
|
|
15
|
+
[](https://github.com/theihtisham/agent-shadow-brain/network)
|
|
16
|
+
[](https://github.com/theihtisham/agent-shadow-brain/issues)
|
|
17
|
+
[](https://github.com/theihtisham/agent-shadow-brain/commits)
|
|
18
|
+
[](https://github.com/theihtisham/agent-shadow-brain)
|
|
19
|
+
[](https://arxiv.org/abs/2504.17033)
|
|
20
|
+
[](https://openreview.net/forum?id=TbqSEUXWaO)
|
|
21
|
+
[](#)
|
|
22
|
+
[](#)
|
|
23
|
+
[](#)
|
|
24
|
+
[](#)
|
|
25
|
+
[](#-architecture)
|
|
26
|
+
[](#-cli-reference)
|
|
27
|
+
[](#)
|
|
28
|
+
[](#-1-singleton-global-brain)
|
|
29
|
+
[](#-2-subconscious-engine)
|
|
30
|
+
[](#-4-l0-sub-millisecond-cache)
|
|
31
|
+
[](#-3-universal-bootstrap-attach-all)
|
|
32
|
+
[](#)
|
|
33
|
+
[](#)
|
|
34
|
+
[](https://github.com/theihtisham/agent-shadow-brain/pulls)
|
|
35
|
+
[](#)
|
|
36
|
+
[](https://github.com/theihtisham)
|
|
37
|
+
|
|
38
|
+
<br/>
|
|
39
|
+
|
|
40
|
+
### **One Brain. Every Agent. Every Project. Every Session. Forever.**
|
|
41
|
+
|
|
42
|
+
**Agent Shadow Brain** is the singleton intelligence layer that makes **every AI coding agent** smarter — Claude Code, Cursor, Cline, Codex, GitHub Copilot, Windsurf, Kilo Code, Roo Code, OpenCode, Aider — all sharing the same brain. What one agent learns, every agent knows. What you discover in project A is available in project B. Forever.
|
|
43
|
+
|
|
44
|
+
> **v5.2.0 — Subconscious Singularity Edition**: One command — `npx @theihtisham/agent-shadow-brain attach-all` — auto-detects every AI agent installed on your machine and wires them to a singleton global brain. Every new session, every agent, every project: Shadow Brain proactively injects the most relevant context **before the agent asks**. Your agents stop forgetting. They stop hallucinating. They start collaborating across tools.
|
|
45
|
+
|
|
46
|
+
```bash
|
|
47
|
+
# 30-second install. Every agent on your machine becomes smarter.
|
|
48
|
+
npx @theihtisham/agent-shadow-brain attach-all
|
|
49
|
+
```
|
|
50
|
+
|
|
51
|
+
[**30-Second Install**](#-getting-started-30-seconds) · [**The Singularity**](#-v520--what-makes-it-singular) · [**Live Dashboard**](#-web-dashboard) · [**Algorithms**](#-algorithms--research) · [**70+ CLI Commands**](#-cli-reference) · [**Architecture**](#-architecture) · [**Star on GitHub ⭐**](https://github.com/theihtisham/agent-shadow-brain)
|
|
52
|
+
|
|
53
|
+
</div>
|
|
54
|
+
|
|
55
|
+
---
|
|
56
|
+
|
|
57
|
+
## ⚡ Why Shadow Brain Exists
|
|
58
|
+
|
|
59
|
+
| Without Shadow Brain | With Shadow Brain |
|
|
60
|
+
|---|---|
|
|
61
|
+
| Every new session, your agent starts from zero | Agent gets a 2K-token briefing of relevant context **before it asks** |
|
|
62
|
+
| Cursor learns something — Claude doesn't know | One singleton brain — what one agent learns, all agents know |
|
|
63
|
+
| Discoveries in project A don't help project B | Cross-project memory federation by default |
|
|
64
|
+
| Each AI tool needs its own setup | One command attaches every agent |
|
|
65
|
+
| Agents hallucinate APIs that don't exist | Adversarial defense + multi-agent consensus voting |
|
|
66
|
+
| Your codebase patterns are reinvented daily | Code DNA fingerprinting enforces your style across all agents |
|
|
67
|
+
| 30 minutes of config before anything works | Zero config, auto-detect, instant value |
|
|
68
|
+
|
|
69
|
+
## 🚀 v5.2.0 — What Makes It Singular
|
|
70
|
+
|
|
71
|
+
### 🧠 1. Singleton Global Brain
|
|
72
|
+
|
|
73
|
+
One database (`~/.shadow-brain/global.db`) is the **source of truth** for every project and every agent on your machine. Project-scoped views give each project its own namespace, but the data lives in one place. SQLite WAL mode + write queue handle concurrent multi-agent writes safely. Auto-vacuum at 500MB, auto-prune at 1GB.
|
|
74
|
+
|
|
75
|
+
```bash
|
|
76
|
+
shadow-brain global stats # See cross-project, cross-agent stats
|
|
77
|
+
shadow-brain global recall # Search across every project + every agent
|
|
78
|
+
```
|
|
79
|
+
|
|
80
|
+
### 💭 2. Subconscious Engine
|
|
81
|
+
|
|
82
|
+
Every time any agent starts a new session, the Subconscious **proactively injects** a tight briefing of:
|
|
83
|
+
- Recent decisions made in this project (last 24h)
|
|
84
|
+
- Active tasks (TODOs, in-progress items)
|
|
85
|
+
- Similar past work from other projects
|
|
86
|
+
- Cross-agent insights (what Claude learned that Cursor should know)
|
|
87
|
+
- Warnings and known pitfalls
|
|
88
|
+
|
|
89
|
+
Hard-capped at 2K tokens. Relevance-ranked. The agent doesn't have to ask — the brain surfaces it implicitly. Like human intuition.
|
|
90
|
+
|
|
91
|
+
```bash
|
|
92
|
+
shadow-brain subconscious inject # See what your agent gets
|
|
93
|
+
shadow-brain subconscious configure --budget 3000 # Tune the budget
|
|
94
|
+
```
|
|
95
|
+
|
|
96
|
+
### 🪝 3. Universal Bootstrap (`attach-all`)
|
|
97
|
+
|
|
98
|
+
One command. Detects every AI agent installed on your machine. Installs the right hook for each one's native mechanism:
|
|
99
|
+
|
|
100
|
+
| Agent | Hook Mechanism |
|
|
101
|
+
|---|---|
|
|
102
|
+
| **Claude Code** | `~/.claude/settings.json` → `hooks.SessionStart` |
|
|
103
|
+
| **Cursor** | `.cursor/rules/shadow-brain.md` + `.cursor/mcp.json` |
|
|
104
|
+
| **Cline** | `~/.vscode/.../cline_mcp_settings.json` |
|
|
105
|
+
| **Windsurf** | `.windsurfrules` + `~/.windsurf/mcp.json` |
|
|
106
|
+
| **Codex** | `~/.codex/config.json` |
|
|
107
|
+
| **Kilo Code** | `~/.kilocode/settings.json` |
|
|
108
|
+
| **OpenCode** | `~/.opencode/settings.json` |
|
|
109
|
+
| **Roo Code** | `~/.roocode/mcp.json` |
|
|
110
|
+
| **Aider** | `~/.aider.conf.yml` `read:` directive |
|
|
111
|
+
| **GitHub Copilot** | `.github/copilot-instructions.md` |
|
|
112
|
+
|
|
113
|
+
```bash
|
|
114
|
+
shadow-brain attach-all # Wire every detected agent
|
|
115
|
+
shadow-brain attach-all --dry-run # See what would be wired
|
|
116
|
+
```
|
|
117
|
+
|
|
118
|
+
### ⚡ 4. L0 Sub-Millisecond Cache
|
|
119
|
+
|
|
120
|
+
In-memory hot tier sits in front of the singleton brain. Hot recalls return in **<1ms**. Byte-budgeted (default 64MB), LRU-evicted, per-namespace. Tracks every access in nanoseconds.
|
|
121
|
+
|
|
122
|
+
```bash
|
|
123
|
+
shadow-brain global cache # Hit rate, top keys, ns-latency
|
|
124
|
+
```
|
|
125
|
+
|
|
126
|
+
### 🎯 5. First-Class Adapters for Every Major Agent
|
|
127
|
+
|
|
128
|
+
Cursor, Windsurf, and Copilot were stub aliases in v5.1.1. In v5.2.0 they each have **dedicated adapters** matching their actual file layouts, conversation parsers, and injection mechanisms.
|
|
129
|
+
|
|
130
|
+
---
|
|
131
|
+
|
|
132
|
+
---
|
|
133
|
+
|
|
134
|
+
## The Problem
|
|
135
|
+
|
|
136
|
+
You use AI coding agents. They're powerful, but:
|
|
137
|
+
|
|
138
|
+
| Problem | Impact |
|
|
139
|
+
|---|---|
|
|
140
|
+
| Agents **forget everything** between sessions | Same mistakes, zero learning |
|
|
141
|
+
| Multiple agents **don't communicate** | Claude doesn't know what Cursor learned |
|
|
142
|
+
| Generated code has **hallucinations** | LLMs fabricate APIs that don't exist |
|
|
143
|
+
| No **cross-session memory** | Every conversation starts from zero |
|
|
144
|
+
| No **cross-project learning** | Discoveries in one project don't help others |
|
|
145
|
+
| Setup is **complicated** | 30 minutes of config before anything works |
|
|
146
|
+
| Each tool needs **separate config** | Claude, Cursor, Cline all need different setup |
|
|
147
|
+
|
|
148
|
+
## The Solution
|
|
149
|
+
|
|
150
|
+
**Shadow Brain v5.1.1** — zero-config, auto-detecting, infinite-intelligence shadow layer:
|
|
151
|
+
|
|
152
|
+
```
|
|
153
|
+
┌──────────────┐ auto-detect ┌──────────────────────────────────────────┐
|
|
154
|
+
│ AI Agents │ ◀──────────────── │ SHADOW BRAIN v5.1.1 │
|
|
155
|
+
│ Claude Code │ MCP auto-install │ │
|
|
156
|
+
│ Cursor │ zero-config │ ┌─ HYPER-COGNITIVE (v5.1.1 NEW) ────┐ │
|
|
157
|
+
│ Kilo Code │ │ │ Built-in LSP Server (stdio+TCP) │ │
|
|
158
|
+
│ Cline │ watches │ │ Custom LLM Fine-Tuning Engine │ │
|
|
159
|
+
│ OpenCode │ ──────────────▶ │ │ Smart Cache (hot/warm/cold tiers) │ │
|
|
160
|
+
│ Codex │ file changes │ │ Intent Engine (NLP commands) │ │
|
|
161
|
+
│ Roo Code │ git commits │ │ Code DNA (9-gene fingerprinting) │ │
|
|
162
|
+
│ Aider │ activity │ │ Temporal Intelligence (velocity) │ │
|
|
163
|
+
│ Windsurf │ │ └────────────────────────────────────┘ │
|
|
164
|
+
└──────────────┘ │ ┌─ ZERO-CONFIG (v5.0.1) ─────────────┐ │
|
|
165
|
+
▲ │ │ Auto-detect + MCP + Dashboard │ │
|
|
166
|
+
│ │ │ Plugins + Export/Import + Hooks │ │
|
|
167
|
+
│ │ └────────────────────────────────────┘ │
|
|
168
|
+
│ │ ┌─ INFINITE MEMORY (v5.0.0) ────────┐ │
|
|
169
|
+
│ │ │ 4-tier hierarchical compression │ │
|
|
170
|
+
│ │ │ Context-triggered recall │ │
|
|
171
|
+
│ │ │ Multi-agent consensus + trust │ │
|
|
172
|
+
│ │ │ Collective cross-project learning │ │
|
|
173
|
+
│ │ └────────────────────────────────────┘ │
|
|
174
|
+
│ │ ┌─ HYPER-INTELLIGENCE (v4) ──────────┐ │
|
|
175
|
+
│ │ │ TurboQuant 6x Compression │ │
|
|
176
|
+
│ │ │ SSSP BMSSP Quantum Routing │ │
|
|
177
|
+
│ │ │ Self-Evolution + CAIP + Swarm │ │
|
|
178
|
+
│ │ │ PageRank + Predictive + Adversarial │ │
|
|
179
|
+
└──────────────────────────┴──────────────────────────────────────────┘
|
|
180
|
+
```
|
|
181
|
+
|
|
182
|
+
---
|
|
183
|
+
|
|
184
|
+
## v5.1.1 — What's New
|
|
185
|
+
|
|
186
|
+
### 6 Hyper-Cognitive Brain Modules
|
|
187
|
+
|
|
188
|
+
<table>
|
|
189
|
+
<tr>
|
|
190
|
+
<td width="50%">
|
|
191
|
+
|
|
192
|
+
### Built-in LSP Server
|
|
193
|
+
|
|
194
|
+
**Language Server Protocol** — real-time IDE feedback without external tools.
|
|
195
|
+
|
|
196
|
+
Pure Node.js LSP server supporting stdio and TCP transport modes:
|
|
197
|
+
- **Diagnostics**: pushes brain insights as IDE warnings/errors
|
|
198
|
+
- **Hover**: shows code health info on hover
|
|
199
|
+
- **Code Actions**: quick-fix suggestions from brain analysis
|
|
200
|
+
- **Zero dependency**: no external LSP libraries required
|
|
201
|
+
|
|
202
|
+
```bash
|
|
203
|
+
shadow-brain lsp --stdio # stdio mode (for editors)
|
|
204
|
+
shadow-brain lsp --tcp 2087 # TCP mode (for remote)
|
|
205
|
+
```
|
|
206
|
+
|
|
207
|
+
### Custom LLM Fine-Tuning Engine
|
|
208
|
+
|
|
209
|
+
**Train on your codebase** — your brain learns YOUR patterns.
|
|
210
|
+
|
|
211
|
+
Automatic training data generation from code changes, insights, and fixes:
|
|
212
|
+
- **Pair extraction**: before/after code → training pairs
|
|
213
|
+
- **Pattern mining**: recurring fix patterns → fine-tune examples
|
|
214
|
+
- **JSONL export**: ready for OpenAI, Anthropic, or local fine-tuning
|
|
215
|
+
- **Quality scoring**: only high-confidence pairs enter the dataset
|
|
216
|
+
|
|
217
|
+
```bash
|
|
218
|
+
shadow-brain fine-tune status # Training data stats
|
|
219
|
+
shadow-brain fine-tune generate [dir] # Generate training pairs
|
|
220
|
+
shadow-brain fine-tune export [path] # Export JSONL dataset
|
|
221
|
+
```
|
|
222
|
+
|
|
223
|
+
</td>
|
|
224
|
+
<td width="50%">
|
|
225
|
+
|
|
226
|
+
### Smart Cache (Multi-Tier)
|
|
227
|
+
|
|
228
|
+
**Hot/Warm/Cold** — intelligent caching with predictive prefetch.
|
|
229
|
+
|
|
230
|
+
Three-tier LRU cache with automatic promotion and demotion:
|
|
231
|
+
- **Hot tier**: 200 entries, fastest access
|
|
232
|
+
- **Warm tier**: 1000 entries, promoted on repeated access
|
|
233
|
+
- **Cold tier**: 5000 entries, compressed storage
|
|
234
|
+
- **Predictive prefetch**: co-access graph predicts what you'll need next
|
|
235
|
+
- **Tag-based invalidation**: invalidate by category
|
|
236
|
+
|
|
237
|
+
```bash
|
|
238
|
+
shadow-brain cache stats # Hit rate, tier sizes, memory usage
|
|
239
|
+
shadow-brain cache clear # Clear all tiers
|
|
240
|
+
```
|
|
241
|
+
|
|
242
|
+
### Intent Engine
|
|
243
|
+
|
|
244
|
+
**Natural language → brain actions** — understand what you mean.
|
|
245
|
+
|
|
246
|
+
Multi-strategy NLP command understanding:
|
|
247
|
+
- **Keyword matching**: fast path for common commands
|
|
248
|
+
- **Fuzzy matching**: handles typos and variations
|
|
249
|
+
- **Context-aware**: uses current file/project context
|
|
250
|
+
- **Confidence scoring**: routes to best-matching action
|
|
251
|
+
|
|
252
|
+
```bash
|
|
253
|
+
shadow-brain intent "find security bugs" # Maps to: analyze + security filter
|
|
254
|
+
shadow-brain intent "show hot files" # Maps to: temporal + heatmap
|
|
255
|
+
```
|
|
256
|
+
|
|
257
|
+
</td>
|
|
258
|
+
</tr>
|
|
259
|
+
<tr>
|
|
260
|
+
<td width="50%">
|
|
261
|
+
|
|
262
|
+
### Code DNA Fingerprinting
|
|
263
|
+
|
|
264
|
+
**9-gene analysis** — every file gets a unique genetic profile.
|
|
265
|
+
|
|
266
|
+
Analyzes code across 9 gene categories to create a structural fingerprint:
|
|
267
|
+
- **Structure genes**: classes, functions, nesting depth
|
|
268
|
+
- **Complexity genes**: cyclomatic complexity, cognitive load
|
|
269
|
+
- **Style genes**: naming conventions, formatting patterns
|
|
270
|
+
- **Dependency genes**: import graph, coupling metrics
|
|
271
|
+
- **Evolution genes**: change frequency, stability score
|
|
272
|
+
- **Similarity matching**: find structurally similar files across projects
|
|
273
|
+
|
|
274
|
+
```bash
|
|
275
|
+
shadow-brain dna analyze [dir] # Generate DNA profiles
|
|
276
|
+
shadow-brain dna compare f1 f2 # Compare two file fingerprints
|
|
277
|
+
shadow-brain dna similar [file] # Find similar files
|
|
278
|
+
```
|
|
279
|
+
|
|
280
|
+
</td>
|
|
281
|
+
<td width="50%">
|
|
282
|
+
|
|
283
|
+
### Temporal Intelligence
|
|
284
|
+
|
|
285
|
+
**Time-series code evolution** — velocity tracking & bug prediction.
|
|
286
|
+
|
|
287
|
+
Continuous monitoring of code change patterns over time:
|
|
288
|
+
- **Velocity metrics**: daily/weekly/monthly rates with trend detection
|
|
289
|
+
- **File heatmaps**: hotness scoring, churn risk, stability
|
|
290
|
+
- **Bug prediction**: 5-factor probability model per file
|
|
291
|
+
- **Anomaly detection**: burst detection, drought alerts, unusual hours
|
|
292
|
+
- **Peak analysis**: identifies your most productive hours and days
|
|
293
|
+
|
|
294
|
+
```bash
|
|
295
|
+
shadow-brain temporal stats # Velocity, anomalies, predictions
|
|
296
|
+
shadow-brain temporal hot # Top 10 hottest files
|
|
297
|
+
shadow-brain temporal predict # Bug risk predictions
|
|
298
|
+
shadow-brain temporal timeline # Recent event timeline
|
|
299
|
+
```
|
|
300
|
+
|
|
301
|
+
</td>
|
|
302
|
+
</tr>
|
|
303
|
+
</table>
|
|
304
|
+
|
|
305
|
+
---
|
|
306
|
+
|
|
307
|
+
## v5.0.1 — What's New
|
|
308
|
+
|
|
309
|
+
### Zero-Config Auto-Setup
|
|
310
|
+
|
|
311
|
+
Install it. That's it. Shadow Brain automatically:
|
|
312
|
+
|
|
313
|
+
- Detects your project type (frontend, backend, fullstack, mobile, game, systems, data-science)
|
|
314
|
+
- Detects 25+ languages, 15+ frameworks, 5+ build tools, 5+ test frameworks
|
|
315
|
+
- Detects which AI tools are installed (Claude Code, Cursor, Kilo Code, Cline, OpenCode, Codex, Aider, Windsurf)
|
|
316
|
+
- Auto-installs MCP server configuration for every detected AI tool
|
|
317
|
+
- Auto-installs git pre-commit and pre-push hooks
|
|
318
|
+
- Generates an optimal brain config based on project complexity
|
|
319
|
+
|
|
320
|
+
```bash
|
|
321
|
+
npm install @theihtisham/agent-shadow-brain
|
|
322
|
+
# → Shadow Brain v5.1.1 auto-configured for my-app (fullstack)
|
|
323
|
+
# → AI tools detected: Claude Code, Cursor
|
|
324
|
+
# → Run: shadow-brain start
|
|
325
|
+
```
|
|
326
|
+
|
|
327
|
+
### MCP Server — Works with EVERY AI Tool
|
|
328
|
+
|
|
329
|
+
The brain exposes itself as a **Model Context Protocol (MCP) server** with 19 tools that any AI tool can call:
|
|
330
|
+
|
|
331
|
+
| MCP Tool | What It Does |
|
|
332
|
+
|---|---|
|
|
333
|
+
| `brain_status` | Get full brain status |
|
|
334
|
+
| `brain_analyze` | Trigger code analysis |
|
|
335
|
+
| `brain_health` | Get health score (A-F) |
|
|
336
|
+
| `brain_insights` | Get recent insights |
|
|
337
|
+
| `brain_fixes` | Get smart fix suggestions |
|
|
338
|
+
| `brain_memory_store` | Store knowledge in hierarchical memory |
|
|
339
|
+
| `brain_memory_search` | Search across all memory tiers |
|
|
340
|
+
| `brain_memory_stats` | Memory compression statistics |
|
|
341
|
+
| `brain_recall` | Context-triggered associative recall |
|
|
342
|
+
| `brain_consensus_propose` | Submit a proposal for agent consensus |
|
|
343
|
+
| `brain_consensus_vote` | Vote on a pending proposal |
|
|
344
|
+
| `brain_collective_propose` | Propose a cross-project rule |
|
|
345
|
+
| `brain_collective_search` | Search collective learning rules |
|
|
346
|
+
| `brain_turbo_search` | Semantic search in compressed memory |
|
|
347
|
+
| `brain_predict` | Predict bug risk for files |
|
|
348
|
+
| `brain_graph_query` | Query knowledge graph |
|
|
349
|
+
| `brain_defense_scan` | Scan for hallucination patterns |
|
|
350
|
+
| `brain_evolve_status` | Get self-evolution status |
|
|
351
|
+
| `brain_dashboard_url` | Get dashboard URL |
|
|
352
|
+
|
|
353
|
+
Works with **Claude Code**, **Cursor**, **Kilo Code**, **Cline**, **OpenCode**, and any MCP-compatible AI tool.
|
|
354
|
+
|
|
355
|
+
```bash
|
|
356
|
+
# Start MCP server (auto-started by AI tools)
|
|
357
|
+
shadow-brain mcp
|
|
358
|
+
|
|
359
|
+
# Cursor compatibility (root path POST + /v1/chat)
|
|
360
|
+
# Claude Code compatibility (JSON-RPC 2.0 over stdio/HTTP)
|
|
361
|
+
```
|
|
362
|
+
|
|
363
|
+
### Rich Web Dashboard
|
|
364
|
+
|
|
365
|
+
Real-time browser dashboard at `http://localhost:7341` with WebSocket live updates:
|
|
366
|
+
|
|
367
|
+
- **Stats Topbar**: 8 live counters — Insights, Fixes, Memory KB, Modules, AI Tools, Evolution Gen, Swarm Convergence, Turbo Compression
|
|
368
|
+
- **Health Score**: A-F grading with dimension breakdown (Security, Quality, Performance, Architecture, Maintainability, Test Coverage)
|
|
369
|
+
- **Memory Tiers**: Visual bars showing Raw (50K), Summary (10K), Pattern (2K), Principle (500) with Turbo compression ratio
|
|
370
|
+
- **Live Insights Stream**: Real-time feed of brain insights with priority badges
|
|
371
|
+
- **Smart Fixes**: Actionable fix suggestions with file/line references
|
|
372
|
+
- **AI Tools Panel**: 8 tools showing connected/detected status
|
|
373
|
+
- **Brain Modules**: 13 modules with active/idle/error indicators
|
|
374
|
+
- **Control Buttons**: Trigger analysis on-demand, refresh data
|
|
375
|
+
|
|
376
|
+
```bash
|
|
377
|
+
shadow-brain start . # Dashboard auto-starts at http://localhost:7341
|
|
378
|
+
shadow-brain dash . # Dashboard only mode
|
|
379
|
+
```
|
|
380
|
+
|
|
381
|
+
### Natural Language Queries
|
|
382
|
+
|
|
383
|
+
Ask your brain questions in plain English:
|
|
384
|
+
|
|
385
|
+
```bash
|
|
386
|
+
shadow-brain ask "What security issues exist in the auth module?"
|
|
387
|
+
shadow-brain ask "Which files have the highest bug risk?"
|
|
388
|
+
shadow-brain ask "What patterns have we learned about React hooks?"
|
|
389
|
+
```
|
|
390
|
+
|
|
391
|
+
### Brain Export/Import — Team Sync
|
|
392
|
+
|
|
393
|
+
Export your brain state and share it with your team:
|
|
394
|
+
|
|
395
|
+
```bash
|
|
396
|
+
shadow-brain export # Full export with all modules
|
|
397
|
+
shadow-brain export -o ./team-brain # Custom output directory
|
|
398
|
+
shadow-brain import brain-export.json # Import + merge
|
|
399
|
+
shadow-brain import brain-export.json --merge --skip swarm,evolution
|
|
400
|
+
```
|
|
401
|
+
|
|
402
|
+
### Plugin System
|
|
403
|
+
|
|
404
|
+
Extend Shadow Brain with custom plugins:
|
|
405
|
+
|
|
406
|
+
```bash
|
|
407
|
+
shadow-brain plugin list # List installed plugins
|
|
408
|
+
shadow-brain plugin create my-plugin # Create from template
|
|
409
|
+
shadow-brain plugin stats # Plugin system statistics
|
|
410
|
+
```
|
|
411
|
+
|
|
412
|
+
Plugins can hook into `pre-analysis` and `post-analysis` pipelines to add custom insights, filter results, or integrate with external services.
|
|
413
|
+
|
|
414
|
+
### 5 New Commands
|
|
415
|
+
|
|
416
|
+
```bash
|
|
417
|
+
shadow-brain off # Clean shutdown
|
|
418
|
+
shadow-brain ask "..." # Natural language query
|
|
419
|
+
shadow-brain export # Export brain state
|
|
420
|
+
shadow-brain import FILE # Import brain state
|
|
421
|
+
shadow-brain plugin CMD # Plugin management
|
|
422
|
+
```
|
|
423
|
+
|
|
424
|
+
---
|
|
425
|
+
|
|
426
|
+
## Getting Started — 30 Seconds
|
|
427
|
+
|
|
428
|
+
```bash
|
|
429
|
+
# Install globally
|
|
430
|
+
npm install -g @theihtisham/agent-shadow-brain
|
|
431
|
+
|
|
432
|
+
# Or use npx (no install needed)
|
|
433
|
+
npx @theihtisham/agent-shadow-brain start .
|
|
434
|
+
```
|
|
435
|
+
|
|
436
|
+
That's literally it. Shadow Brain will:
|
|
437
|
+
1. **Auto-detect** your project, languages, frameworks, AI tools (zero config)
|
|
438
|
+
2. **Auto-install** MCP server config for Claude Code, Cursor, Kilo Code, Cline, etc.
|
|
439
|
+
3. **Auto-install** git hooks for pre-commit and pre-push
|
|
440
|
+
4. Watch your project files for changes
|
|
441
|
+
5. Analyze every change with LLM + 22 brain modules
|
|
442
|
+
6. Inject expert insights into your agent's memory
|
|
443
|
+
7. Store all knowledge in hierarchical memory for infinite retention
|
|
444
|
+
8. Activate past memories automatically based on context
|
|
445
|
+
9. Share verified patterns across projects via collective learning
|
|
446
|
+
10. Compress knowledge with TurboQuant for infinite retention
|
|
447
|
+
11. Evolve its own rules via genetic algorithm
|
|
448
|
+
12. Share insights across agents via CAIP
|
|
449
|
+
13. Detect hallucinations with adversarial verification
|
|
450
|
+
14. Run consensus when multiple agents disagree
|
|
451
|
+
15. Serve a rich web dashboard at `localhost:7341`
|
|
452
|
+
|
|
453
|
+
### Quick Review (No Watch Mode)
|
|
454
|
+
|
|
455
|
+
```bash
|
|
456
|
+
shadow-brain review . # One-shot analysis
|
|
457
|
+
shadow-brain review . --show-health # + health score
|
|
458
|
+
shadow-brain review . --show-fixes # + fix suggestions
|
|
459
|
+
shadow-brain review . --output json # JSON for scripting
|
|
460
|
+
```
|
|
461
|
+
|
|
462
|
+
---
|
|
463
|
+
|
|
464
|
+
## Web Dashboard
|
|
465
|
+
|
|
466
|
+
The v5.1.1 dashboard provides a real-time view into your brain:
|
|
467
|
+
|
|
468
|
+
```
|
|
469
|
+
┌──────────────────────────────────────────────────────────────────────────┐
|
|
470
|
+
│ Shadow Brain v5.1.1 [Analyze] [↻] │
|
|
471
|
+
├──────────────────────────────────────────────────────────────────────────┤
|
|
472
|
+
│ Insights: 142 │ Fixes: 38 │ Memory: 247KB │ Modules: 22/22 │ ... │
|
|
473
|
+
├──────────────┬─────────────────────────────────────┬───────────────────┤
|
|
474
|
+
│ Health: 87% │ Live Insights Stream │ AI Tools │
|
|
475
|
+
│ ████████░░ │ ● [HIGH] SQL injection in login.ts │ ● Claude Code ✓ │
|
|
476
|
+
│ Security: A │ ● [MED] Missing error handling │ ● Cursor ✓ │
|
|
477
|
+
│ Quality: B+ │ ● [LOW] Unused import in utils │ ● Kilo Code ✓ │
|
|
478
|
+
│ Perf: A- │ ● [CRIT] Exposed API key │ ● Cline — │
|
|
479
|
+
│ Arch: B │ │ ● OpenCode — │
|
|
480
|
+
│ Maintain: A ├─────────────────────────────────────┤ ● Codex — │
|
|
481
|
+
│ Tests: C+ │ Smart Fixes │ ● Aider — │
|
|
482
|
+
│ │ 1. Add parameterized query... │ ● Windsurf — │
|
|
483
|
+
├──────────────┤ 2. Wrap in try/catch with... ├───────────────────┤
|
|
484
|
+
│ Memory Tiers │ 3. Remove unused import on L42 │ Brain Modules │
|
|
485
|
+
│ Raw: 8421 │ │ ● Hierarchical ✓ │
|
|
486
|
+
│ Summ: 1203 │ │ ● TurboMemory ✓ │
|
|
487
|
+
│ Patt: 187 │ │ ● Consensus ✓ │
|
|
488
|
+
│ Princ: 24 │ │ ● Swarm ✓ │
|
|
489
|
+
│ Turbo: 6.2x │ │ ● Evolution ✓ │
|
|
490
|
+
│ │ │ ● Adversarial ✓ │
|
|
491
|
+
└──────────────┴─────────────────────────────────────┴───────────────────┘
|
|
492
|
+
```
|
|
493
|
+
|
|
494
|
+
---
|
|
495
|
+
|
|
496
|
+
## v5.0.0 Infinite Intelligence Engine
|
|
497
|
+
|
|
498
|
+
### 4 Breakthrough Modules
|
|
499
|
+
|
|
500
|
+
<table>
|
|
501
|
+
<tr>
|
|
502
|
+
<td width="50%">
|
|
503
|
+
|
|
504
|
+
### Infinite Memory Layer
|
|
505
|
+
|
|
506
|
+
**Hierarchical Memory Compression** — 4-tier pyramid that never forgets.
|
|
507
|
+
|
|
508
|
+
Knowledge flows through compression tiers as it ages:
|
|
509
|
+
- **Raw** (50K entries) — full detail, hot path
|
|
510
|
+
- **Summary** (10K) — condensed key points
|
|
511
|
+
- **Pattern** (2K) — recurring patterns extracted
|
|
512
|
+
- **Principle** (500) — timeless design rules
|
|
513
|
+
|
|
514
|
+
Drill-down and drill-up between tiers. Older knowledge compresses, never deletes.
|
|
515
|
+
|
|
516
|
+
```bash
|
|
517
|
+
shadow-brain memory stats # 4 tiers | 12,847 entries
|
|
518
|
+
shadow-brain memory store "..." # Store new knowledge
|
|
519
|
+
shadow-brain memory search "auth" # Cross-tier semantic search
|
|
520
|
+
```
|
|
521
|
+
|
|
522
|
+
**Context-Triggered Associative Recall** — memories that activate themselves.
|
|
523
|
+
|
|
524
|
+
Instead of requiring explicit search queries, this engine monitors your current work context and automatically surfaces relevant past knowledge:
|
|
525
|
+
- File path pattern matching (editing `auth/*.ts` → security memories)
|
|
526
|
+
- Keyword detection (typing "database" → SQL pattern memories)
|
|
527
|
+
- Category association (bug-fix context → past fix patterns)
|
|
528
|
+
- Co-occurrence networks (memories activated together strengthen their link)
|
|
529
|
+
- Learned trigger patterns that improve with use
|
|
530
|
+
|
|
531
|
+
```bash
|
|
532
|
+
shadow-brain recall . --file src/auth/login.ts --keywords "jwt,session"
|
|
533
|
+
# → Activates 8 memories | Top: "JWT rotation pattern (92%)"
|
|
534
|
+
```
|
|
535
|
+
|
|
536
|
+
</td>
|
|
537
|
+
<td width="50%">
|
|
538
|
+
|
|
539
|
+
### Multi-Agent Intelligence Layer
|
|
540
|
+
|
|
541
|
+
**Consensus Engine** — voting, trust scoring, conflict resolution.
|
|
542
|
+
|
|
543
|
+
When multiple AI agents observe the same codebase, they may produce conflicting insights. This engine:
|
|
544
|
+
1. Collects proposals from agents
|
|
545
|
+
2. Opens voting windows (agree/disagree/abstain with confidence)
|
|
546
|
+
3. Computes trust-weighted agreement scores
|
|
547
|
+
4. Resolves conflicts via confidence intervals
|
|
548
|
+
5. Tracks long-term agent trust based on proposal accuracy
|
|
549
|
+
|
|
550
|
+
```bash
|
|
551
|
+
shadow-brain consensus status
|
|
552
|
+
shadow-brain consensus propose "Use bcrypt for all password hashing" security 0.95
|
|
553
|
+
shadow-brain consensus vote <id> agree 0.9 "Industry standard"
|
|
554
|
+
shadow-brain consensus results
|
|
555
|
+
```
|
|
556
|
+
|
|
557
|
+
**Collective Learning** — discoveries flow across all projects.
|
|
558
|
+
|
|
559
|
+
When you discover a pattern in Project A, it becomes a verified rule available in Project B:
|
|
560
|
+
- Rule proposal with evidence and verification
|
|
561
|
+
- Viral propagation across connected brain instances
|
|
562
|
+
- Accuracy tracking per rule (tested vs. false positive)
|
|
563
|
+
- Category-based organization (security, performance, architecture...)
|
|
564
|
+
- Trust scores for rule sources
|
|
565
|
+
|
|
566
|
+
```bash
|
|
567
|
+
shadow-brain collective status
|
|
568
|
+
shadow-brain collective propose "Always validate file upload MIME types" security
|
|
569
|
+
shadow-brain collective search performance
|
|
570
|
+
```
|
|
571
|
+
|
|
572
|
+
</td>
|
|
573
|
+
</tr>
|
|
574
|
+
</table>
|
|
575
|
+
|
|
576
|
+
---
|
|
577
|
+
|
|
578
|
+
## v4.0.0 Hyper-Intelligence Engine
|
|
579
|
+
|
|
580
|
+
### 8 Breakthrough Modules (included in v5.1.1)
|
|
581
|
+
|
|
582
|
+
<table>
|
|
583
|
+
<tr>
|
|
584
|
+
<td width="50%">
|
|
585
|
+
|
|
586
|
+
### Quantum Memory Layer
|
|
587
|
+
|
|
588
|
+
**TurboQuant Infinite Memory** — never forget anything, ever.
|
|
589
|
+
|
|
590
|
+
Based on Google Research's **TurboQuant** (ICLR 2026):
|
|
591
|
+
- **PolarQuant**: 2 bits per dimension (polar coordinate quantization)
|
|
592
|
+
- **QJL Residual**: 1 bit per dimension (random rotation + sign)
|
|
593
|
+
- **Total**: 3 bits/dim = **6x compression** with <1% accuracy loss
|
|
594
|
+
- **Result**: Infinite retention — old knowledge is compressed, never deleted
|
|
595
|
+
|
|
596
|
+
```bash
|
|
597
|
+
shadow-brain turbo stats
|
|
598
|
+
shadow-brain turbo search "authentication patterns"
|
|
599
|
+
```
|
|
600
|
+
|
|
601
|
+
**SSSP Quantum Router** — deterministic shortest-path message routing.
|
|
602
|
+
|
|
603
|
+
Based on **"Breaking the Sorting Barrier"** (arXiv 2504.17033, Duan et al.):
|
|
604
|
+
- **BMSSP Algorithm**: O(m log<sup>2/3</sup> n) — breaks the sorting barrier
|
|
605
|
+
- **Deterministic**: No randomization, provably correct
|
|
606
|
+
|
|
607
|
+
```bash
|
|
608
|
+
shadow-brain route status
|
|
609
|
+
shadow-brain route find node-A node-B
|
|
610
|
+
```
|
|
611
|
+
|
|
612
|
+
</td>
|
|
613
|
+
<td width="50%">
|
|
614
|
+
|
|
615
|
+
### Self-Evolution Layer
|
|
616
|
+
|
|
617
|
+
**Genetic Algorithm Self-Evolution** — rules that write themselves.
|
|
618
|
+
|
|
619
|
+
- **Population**: 50 genetic rule chromosomes
|
|
620
|
+
- **Selection**: Tournament (k=5) — survival of the fittest
|
|
621
|
+
- **Crossover**: Single-point recombination
|
|
622
|
+
- **Mutation**: Gaussian noise (Box-Muller transform, sigma=0.1)
|
|
623
|
+
- **Fitness**: accuracy x coverage x (1 / falsePositiveRate)
|
|
624
|
+
|
|
625
|
+
```bash
|
|
626
|
+
shadow-brain evolve status
|
|
627
|
+
shadow-brain evolve run
|
|
628
|
+
shadow-brain evolve best-rules security
|
|
629
|
+
```
|
|
630
|
+
|
|
631
|
+
**Cross-Agent Intelligence Protocol (CAIP)** — agents that talk to each other.
|
|
632
|
+
|
|
633
|
+
Your Claude Code session learns what your Cursor session discovered:
|
|
634
|
+
- Zero-config broadcast protocol
|
|
635
|
+
- Agent identification + insight tagging
|
|
636
|
+
- Multi-agent consensus building
|
|
637
|
+
|
|
638
|
+
```bash
|
|
639
|
+
shadow-brain caip status
|
|
640
|
+
shadow-brain caip broadcast "Security pattern detected"
|
|
641
|
+
```
|
|
642
|
+
|
|
643
|
+
</td>
|
|
644
|
+
</tr>
|
|
645
|
+
<tr>
|
|
646
|
+
<td width="50%">
|
|
647
|
+
|
|
648
|
+
### Defense & Trust Layer
|
|
649
|
+
|
|
650
|
+
**Adversarial Hallucination Defense** — trust nothing, verify everything.
|
|
651
|
+
|
|
652
|
+
- Cross-reference every critical insight against actual code
|
|
653
|
+
- Evidence scoring with confidence thresholds
|
|
654
|
+
- Verdict: `real` | `hallucinated` | `uncertain`
|
|
655
|
+
|
|
656
|
+
```bash
|
|
657
|
+
shadow-brain defense status
|
|
658
|
+
shadow-brain defense scan "This API endpoint exists at /api/v2/users"
|
|
659
|
+
```
|
|
660
|
+
|
|
661
|
+
**Swarm Intelligence** — Ant Colony Optimization for file prioritization.
|
|
662
|
+
|
|
663
|
+
- Pheromone deposit: +3 for critical, +2 for high, +1 for medium
|
|
664
|
+
- Evaporation: decay over time to avoid staleness
|
|
665
|
+
|
|
666
|
+
```bash
|
|
667
|
+
shadow-brain swarm status
|
|
668
|
+
shadow-brain swarm priorities
|
|
669
|
+
```
|
|
670
|
+
|
|
671
|
+
</td>
|
|
672
|
+
<td width="50%">
|
|
673
|
+
|
|
674
|
+
### Analytics & Prediction Layer
|
|
675
|
+
|
|
676
|
+
**Knowledge Graph + PageRank** — code impact radar.
|
|
677
|
+
|
|
678
|
+
- Identify high-impact files (change one, break many)
|
|
679
|
+
- Cycle detection in dependency graphs
|
|
680
|
+
- Entity extraction with file/line tracking
|
|
681
|
+
|
|
682
|
+
```bash
|
|
683
|
+
shadow-brain graph build .
|
|
684
|
+
shadow-brain graph pagerank
|
|
685
|
+
shadow-brain graph cycles
|
|
686
|
+
```
|
|
687
|
+
|
|
688
|
+
**Predictive Engine** — predict bugs before they happen.
|
|
689
|
+
|
|
690
|
+
- Bug risk scoring: `low` | `medium` | `high` | `critical`
|
|
691
|
+
- Technical debt forecasting
|
|
692
|
+
- Anomaly detection in change patterns
|
|
693
|
+
|
|
694
|
+
```bash
|
|
695
|
+
shadow-brain predict bugs .
|
|
696
|
+
```
|
|
697
|
+
|
|
698
|
+
</td>
|
|
699
|
+
</tr>
|
|
700
|
+
</table>
|
|
701
|
+
|
|
702
|
+
---
|
|
703
|
+
|
|
704
|
+
## Algorithms & Research
|
|
705
|
+
|
|
706
|
+
Shadow Brain v5.1.1 implements algorithms from **peer-reviewed research**:
|
|
707
|
+
|
|
708
|
+
| Algorithm | Paper / Origin | Application | Complexity |
|
|
709
|
+
|---|---|---|---|
|
|
710
|
+
| **Hierarchical Compression** | Novel 4-tier system | Infinite memory with drill-down | O(n) per tier |
|
|
711
|
+
| **Associative Recall** | Spreading activation model | Context-triggered memory activation | O(n x k) |
|
|
712
|
+
| **Consensus Protocol** | Byzantine fault tolerance | Multi-agent agreement with trust | O(v x log v) |
|
|
713
|
+
| **TurboQuant** | Google Research, ICLR 2026 | Vector compression (PolarQuant + QJL) | O(n) per vector |
|
|
714
|
+
| **BMSSP (SSSP)** | arXiv 2504.17033, Duan et al. | Neural mesh message routing | O(m log<sup>2/3</sup> n) |
|
|
715
|
+
| **Shannon Entropy** | Claude Shannon, 1948 | Cross-project insight relevance scoring | O(n) |
|
|
716
|
+
| **Bayesian Inference** | Thomas Bayes, 1763 | Confidence updating, meta-learning | O(1) per update |
|
|
717
|
+
| **Cosine Similarity** | Vector space model | Knowledge deduplication | O(d) per pair |
|
|
718
|
+
| **PageRank** | Brin & Page, 1998 | Code entity impact analysis | O(V + E) per iteration |
|
|
719
|
+
| **Tarjan's SCC** | Robert Tarjan, 1972 | Dependency cycle detection | O(V + E) |
|
|
720
|
+
| **Winnowing** | Schleimer et al., 2003 | Code duplicate fingerprinting | O(n) |
|
|
721
|
+
| **Box-Muller** | Box & Muller, 1958 | Gaussian mutation in genetic algorithm | O(1) per sample |
|
|
722
|
+
| **Ant Colony (ACO)** | Dorigo, 1992 | File priority pheromone system | O(n x m) |
|
|
723
|
+
| **Tournament Selection** | Goldberg, 1989 | Genetic rule selection (k=5) | O(k) per selection |
|
|
724
|
+
|
|
725
|
+
### TurboQuant Pipeline Detail
|
|
726
|
+
|
|
727
|
+
```
|
|
728
|
+
Input Vector (64-dim, float64)
|
|
729
|
+
|
|
|
730
|
+
v
|
|
731
|
+
+-------------+
|
|
732
|
+
| PolarQuant |-- Cartesian -> Polar coordinates
|
|
733
|
+
| 2 bits/dim |-- Quantize angles to 4 levels (2 bits each)
|
|
734
|
+
+------+-------+ Pack into Uint8Array
|
|
735
|
+
|
|
|
736
|
+
v
|
|
737
|
+
+-------------+
|
|
738
|
+
| QJL Residual|-- Random rotation (Hadamard-like)
|
|
739
|
+
| 1 bit/dim |-- Sign extraction (1 bit per dim)
|
|
740
|
+
+------+-------+ Pack into Uint8Array
|
|
741
|
+
|
|
|
742
|
+
v
|
|
743
|
+
TurboVector { polar: Uint8Array, qjl: Uint8Array, dim: number, radius: number }
|
|
744
|
+
= 3 bits/dim total = 6x compression from float64
|
|
745
|
+
= ZERO FORGETTING -- compressed knowledge stays searchable forever
|
|
746
|
+
```
|
|
747
|
+
|
|
748
|
+
---
|
|
749
|
+
|
|
750
|
+
## Features
|
|
751
|
+
|
|
752
|
+
<table>
|
|
753
|
+
<tr>
|
|
754
|
+
<td width="20%">
|
|
755
|
+
|
|
756
|
+
### Hyper-Cognitive (v5.1.1)
|
|
757
|
+
- Built-in LSP Server
|
|
758
|
+
- Custom LLM Fine-Tuning
|
|
759
|
+
- Smart Cache (3-tier)
|
|
760
|
+
- Intent Engine (NLP)
|
|
761
|
+
- Code DNA Fingerprinting
|
|
762
|
+
- Temporal Intelligence
|
|
763
|
+
|
|
764
|
+
</td>
|
|
765
|
+
<td width="20%">
|
|
766
|
+
|
|
767
|
+
### Zero Config (v5.0.1)
|
|
768
|
+
- Auto-detect project + tools
|
|
769
|
+
- MCP for ALL AI tools
|
|
770
|
+
- Rich web dashboard
|
|
771
|
+
- Natural language queries
|
|
772
|
+
- Plugin system
|
|
773
|
+
- Brain export/import
|
|
774
|
+
- Git hook auto-install
|
|
775
|
+
|
|
776
|
+
</td>
|
|
777
|
+
<td width="20%">
|
|
778
|
+
|
|
779
|
+
### Infinite Intelligence (v5)
|
|
780
|
+
- 4-tier hierarchical memory
|
|
781
|
+
- Context-triggered recall
|
|
782
|
+
- Multi-agent consensus
|
|
783
|
+
- Collective cross-project learning
|
|
784
|
+
|
|
785
|
+
</td>
|
|
786
|
+
<td width="20%">
|
|
787
|
+
|
|
788
|
+
### Hyper-Intelligence (v4)
|
|
789
|
+
- TurboQuant infinite memory
|
|
790
|
+
- SSSP quantum routing
|
|
791
|
+
- Self-evolving genetic rules
|
|
792
|
+
- Cross-agent protocol (CAIP)
|
|
793
|
+
- Adversarial defense
|
|
794
|
+
- Swarm intelligence (ACO)
|
|
795
|
+
- Knowledge Graph + PageRank
|
|
796
|
+
- Predictive bug forecasting
|
|
797
|
+
|
|
798
|
+
</td>
|
|
799
|
+
<td width="20%">
|
|
800
|
+
|
|
801
|
+
### Developer Experience
|
|
802
|
+
- **60+ CLI commands**
|
|
803
|
+
- MCP Server (19 tools)
|
|
804
|
+
- Built-in LSP Server
|
|
805
|
+
- Web dashboard (real-time)
|
|
806
|
+
- Terminal UI (Ink/React)
|
|
807
|
+
- Pre-commit hooks
|
|
808
|
+
- GitHub Actions CI
|
|
809
|
+
- Plugin architecture
|
|
810
|
+
|
|
811
|
+
</td>
|
|
812
|
+
</tr>
|
|
813
|
+
</table>
|
|
814
|
+
|
|
815
|
+
---
|
|
816
|
+
|
|
817
|
+
## Architecture
|
|
818
|
+
|
|
819
|
+
### v5.1.1 Module Map (22 modules)
|
|
820
|
+
|
|
821
|
+
| Module | File | What It Does | Since |
|
|
822
|
+
|---|---|---|---|
|
|
823
|
+
| **LSP Server** | `brain/lsp-server.ts` | Built-in Language Server Protocol (stdio+TCP) | v5.1.1 |
|
|
824
|
+
| **Fine-Tuning Engine** | `brain/fine-tuning-engine.ts` | Custom LLM training data generation | v5.1.1 |
|
|
825
|
+
| **Smart Cache** | `brain/smart-cache.ts` | Multi-tier caching with predictive prefetch | v5.1.1 |
|
|
826
|
+
| **Intent Engine** | `brain/intent-engine.ts` | Natural language command understanding | v5.1.1 |
|
|
827
|
+
| **Code DNA** | `brain/code-dna.ts` | 9-gene code fingerprinting + similarity | v5.1.1 |
|
|
828
|
+
| **Temporal Intelligence** | `brain/temporal-intelligence.ts` | Velocity tracking, heatmaps, bug prediction | v5.1.1 |
|
|
829
|
+
| **Auto Setup** | `brain/auto-setup.ts` | Zero-config project + tool detection | v5.0.1 |
|
|
830
|
+
| **MCP Server** | `brain/mcp-server.ts` | 19-tool MCP server for all AI tools | v5.0.1 |
|
|
831
|
+
| **Plugin System** | `brain/plugin-system.ts` | Hookable analysis pipeline | v5.0.1 |
|
|
832
|
+
| **Brain Portability** | `brain/brain-portability.ts` | Export/import brain state | v5.0.1 |
|
|
833
|
+
| **Dashboard** | `dashboard/server.ts` | Real-time WebSocket dashboard | v5.0.1 |
|
|
834
|
+
| **Hierarchical Memory** | `brain/hierarchical-memory.ts` | 4-tier infinite memory | v5.0.0 |
|
|
835
|
+
| **Context Recall** | `brain/context-recall.ts` | Associative memory activation | v5.0.0 |
|
|
836
|
+
| **Consensus Engine** | `brain/consensus-engine.ts` | Multi-agent voting + trust | v5.0.0 |
|
|
837
|
+
| **Collective Learning** | `brain/collective-learning.ts` | Cross-project rule sharing | v5.0.0 |
|
|
838
|
+
| **TurboMemory** | `brain/turbo-memory.ts` | 6x compressed infinite memory | v4.0.0 |
|
|
839
|
+
| **SSSP Router** | `brain/sssp-router.ts` | Sub-sorting routing | v4.0.0 |
|
|
840
|
+
| **Self-Evolution** | `brain/self-evolution.ts` | Genetic rule optimization | v4.0.0 |
|
|
841
|
+
| **Cross-Agent** | `brain/cross-agent-protocol.ts` | Multi-agent broadcast | v4.0.0 |
|
|
842
|
+
| **Adversarial** | `brain/adversarial-defense.ts` | Hallucination detection | v4.0.0 |
|
|
843
|
+
| **Swarm** | `brain/swarm-intelligence.ts` | File priority pheromones | v4.0.0 |
|
|
844
|
+
| **Knowledge Graph** | `brain/knowledge-graph.ts` | PageRank impact analysis | v4.0.0 |
|
|
845
|
+
| **Predictive** | `brain/predictive-engine.ts` | Bug risk & debt forecasting | v4.0.0 |
|
|
846
|
+
|
|
847
|
+
---
|
|
848
|
+
|
|
849
|
+
## CLI Reference
|
|
850
|
+
|
|
851
|
+
### Core Commands
|
|
852
|
+
|
|
853
|
+
| Command | Description |
|
|
854
|
+
|---|---|
|
|
855
|
+
| `shadow-brain start [dir]` | Start real-time watching + dashboard + MCP |
|
|
856
|
+
| `shadow-brain off` | Clean shutdown |
|
|
857
|
+
| `shadow-brain review [dir]` | One-shot code analysis |
|
|
858
|
+
| `shadow-brain health [dir]` | Health score (A-F grading) |
|
|
859
|
+
| `shadow-brain fix [dir]` | Smart fix suggestions |
|
|
860
|
+
| `shadow-brain report [dir]` | HTML/MD/JSON reports |
|
|
861
|
+
| `shadow-brain metrics [dir]` | Code metrics |
|
|
862
|
+
| `shadow-brain scan [dir]` | Vulnerability scanner |
|
|
863
|
+
| `shadow-brain pr [dir]` | Generate PR description |
|
|
864
|
+
| `shadow-brain commit-msg [dir]` | Generate commit message |
|
|
865
|
+
| `shadow-brain ci [dir]` | GitHub Actions workflow |
|
|
866
|
+
| `shadow-brain hook [dir]` | Pre-commit hook |
|
|
867
|
+
| `shadow-brain dash [dir]` | Web dashboard |
|
|
868
|
+
| `shadow-brain mcp` | MCP server mode |
|
|
869
|
+
| `shadow-brain inject <msg>` | Inject into agent memory |
|
|
870
|
+
| `shadow-brain status` | Current configuration |
|
|
871
|
+
| `shadow-brain setup` | Interactive setup wizard |
|
|
872
|
+
| `shadow-brain doctor` | Health check & diagnostics |
|
|
873
|
+
| `shadow-brain ask <question>` | Natural language brain query |
|
|
874
|
+
|
|
875
|
+
### v5.1.1 Hyper-Cognitive Commands (NEW)
|
|
876
|
+
|
|
877
|
+
| Command | Description |
|
|
878
|
+
|---|---|
|
|
879
|
+
| `shadow-brain lsp [--stdio\|--tcp PORT]` | Start built-in LSP server |
|
|
880
|
+
| `shadow-brain fine-tune status` | Fine-tuning training data stats |
|
|
881
|
+
| `shadow-brain fine-tune generate [dir]` | Generate training pairs from code |
|
|
882
|
+
| `shadow-brain fine-tune export [path]` | Export JSONL dataset for fine-tuning |
|
|
883
|
+
| `shadow-brain cache stats` | Smart cache hit rate, tier sizes, memory |
|
|
884
|
+
| `shadow-brain cache clear` | Clear all cache tiers |
|
|
885
|
+
| `shadow-brain intent <text>` | Parse natural language intent |
|
|
886
|
+
| `shadow-brain dna analyze [dir]` | Generate Code DNA profiles |
|
|
887
|
+
| `shadow-brain dna compare <f1> <f2>` | Compare two file fingerprints |
|
|
888
|
+
| `shadow-brain dna similar <file>` | Find structurally similar files |
|
|
889
|
+
| `shadow-brain temporal stats` | Velocity, anomalies, predictions |
|
|
890
|
+
| `shadow-brain temporal hot` | Top 10 hottest files by churn |
|
|
891
|
+
| `shadow-brain temporal predict` | Bug risk predictions |
|
|
892
|
+
| `shadow-brain temporal timeline` | Recent event timeline |
|
|
893
|
+
|
|
894
|
+
### v5.0.1 Commands
|
|
895
|
+
|
|
896
|
+
| Command | Description |
|
|
897
|
+
|---|---|
|
|
898
|
+
| `shadow-brain off` | Clean shutdown of all brain processes |
|
|
899
|
+
| `shadow-brain ask <question>` | Natural language query to the brain |
|
|
900
|
+
| `shadow-brain export [-o dir]` | Export full brain state to JSON |
|
|
901
|
+
| `shadow-brain import <file> [--merge]` | Import brain state |
|
|
902
|
+
| `shadow-brain plugin list` | List installed plugins |
|
|
903
|
+
| `shadow-brain plugin create <name>` | Create plugin from template |
|
|
904
|
+
| `shadow-brain plugin stats` | Plugin system statistics |
|
|
905
|
+
|
|
906
|
+
### v5.0.0 Infinite Intelligence Commands
|
|
907
|
+
|
|
908
|
+
| Command | Description |
|
|
909
|
+
|---|---|
|
|
910
|
+
| `shadow-brain memory stats` | 4-tier hierarchical memory stats |
|
|
911
|
+
| `shadow-brain memory store <text>` | Store knowledge at raw tier |
|
|
912
|
+
| `shadow-brain memory search <query>` | Cross-tier semantic search |
|
|
913
|
+
| `shadow-brain recall . --file <f>` | Context-triggered associative recall |
|
|
914
|
+
| `shadow-brain consensus status` | Multi-agent consensus stats |
|
|
915
|
+
| `shadow-brain consensus propose <text> <cat> <conf>` | Submit proposal for voting |
|
|
916
|
+
| `shadow-brain consensus vote <id> <vote> <conf> <reason>` | Vote on a proposal |
|
|
917
|
+
| `shadow-brain consensus results` | All resolved proposals |
|
|
918
|
+
| `shadow-brain collective status` | Cross-project rule stats |
|
|
919
|
+
| `shadow-brain collective propose <text> <cat>` | Propose a collective rule |
|
|
920
|
+
| `shadow-brain collective search <cat>` | Search collective rules by category |
|
|
921
|
+
| `shadow-brain v5 [dir]` | Run ALL v5.0.0 analyses |
|
|
922
|
+
|
|
923
|
+
### v4.0.0 Hyper-Intelligence Commands
|
|
924
|
+
|
|
925
|
+
| Command | Description |
|
|
926
|
+
|---|---|
|
|
927
|
+
| `shadow-brain turbo stats` | TurboQuant infinite memory stats |
|
|
928
|
+
| `shadow-brain turbo search <q>` | Semantic search compressed memory |
|
|
929
|
+
| `shadow-brain route status` | SSSP routing status |
|
|
930
|
+
| `shadow-brain route find <from> <to>` | Shortest path query |
|
|
931
|
+
| `shadow-brain caip status` | Cross-agent protocol status |
|
|
932
|
+
| `shadow-brain caip broadcast <msg>` | Broadcast to all agents |
|
|
933
|
+
| `shadow-brain evolve status` | Genetic algorithm generation stats |
|
|
934
|
+
| `shadow-brain evolve run` | Trigger evolution cycle |
|
|
935
|
+
| `shadow-brain evolve best-rules [cat]` | Top evolved rules by category |
|
|
936
|
+
| `shadow-brain predict bugs [dir]` | Bug risk prediction |
|
|
937
|
+
| `shadow-brain graph build [dir]` | Build knowledge graph + PageRank |
|
|
938
|
+
| `shadow-brain swarm status` | Swarm convergence + priorities |
|
|
939
|
+
| `shadow-brain swarm priorities` | File hotspot ranking |
|
|
940
|
+
| `shadow-brain defense status` | Adversarial defense statistics |
|
|
941
|
+
| `shadow-brain defense scan <text>` | Scan text for hallucination patterns |
|
|
942
|
+
| `shadow-brain v4 [dir]` | Run ALL v4.0.0 analyses |
|
|
943
|
+
|
|
944
|
+
---
|
|
945
|
+
|
|
946
|
+
## Programmatic API
|
|
947
|
+
|
|
948
|
+
```typescript
|
|
949
|
+
import {
|
|
950
|
+
Orchestrator,
|
|
951
|
+
// v5.1.1 — Hyper-Cognitive modules
|
|
952
|
+
LSPServer,
|
|
953
|
+
FineTuningEngine,
|
|
954
|
+
SmartCache,
|
|
955
|
+
IntentEngine,
|
|
956
|
+
CodeDNA,
|
|
957
|
+
TemporalIntelligence,
|
|
958
|
+
// v5.0.1 — Zero-Config modules
|
|
959
|
+
AutoSetup,
|
|
960
|
+
PluginSystem,
|
|
961
|
+
BrainPortability,
|
|
962
|
+
// v5.0.0 — Infinite Intelligence modules
|
|
963
|
+
HierarchicalMemory,
|
|
964
|
+
ContextRecall,
|
|
965
|
+
ConsensusEngine,
|
|
966
|
+
CollectiveLearning,
|
|
967
|
+
// v4.0.0 — Hyper-Intelligence modules
|
|
968
|
+
TurboMemory,
|
|
969
|
+
SSSPRouter,
|
|
970
|
+
SelfEvolution,
|
|
971
|
+
CrossAgentProtocol,
|
|
972
|
+
AdversarialDefense,
|
|
973
|
+
SwarmIntelligence,
|
|
974
|
+
KnowledgeGraph,
|
|
975
|
+
PredictiveEngine,
|
|
976
|
+
NeuralMesh,
|
|
977
|
+
} from '@theihtisham/agent-shadow-brain';
|
|
978
|
+
|
|
979
|
+
// v5.1.1 — Built-in LSP server
|
|
980
|
+
const lsp = new LSPServer('./my-project');
|
|
981
|
+
lsp.start({ mode: 'tcp', port: 2087 }); // or { mode: 'stdio' }
|
|
982
|
+
|
|
983
|
+
// v5.1.1 — Smart Cache
|
|
984
|
+
const cache = new SmartCache({ maxHotEntries: 200, enablePrefetch: true });
|
|
985
|
+
cache.set('analysis:auth', analysisResult, { ttl: 300_000, tags: ['security'] });
|
|
986
|
+
const cached = cache.get('analysis:auth');
|
|
987
|
+
|
|
988
|
+
// v5.1.1 — Temporal Intelligence
|
|
989
|
+
const temporal = new TemporalIntelligence('./my-project');
|
|
990
|
+
temporal.recordCommit(['src/auth.ts'], 'fix: JWT rotation');
|
|
991
|
+
const velocity = temporal.getVelocity(); // { daily, weekly, trend: 'accelerating' }
|
|
992
|
+
const bugRisks = temporal.predictBugs(10);
|
|
993
|
+
|
|
994
|
+
// v5.1.1 — Code DNA
|
|
995
|
+
const dna = new CodeDNA('./my-project');
|
|
996
|
+
const profile = await dna.analyze('src/auth.ts');
|
|
997
|
+
const similar = await dna.findSimilar('src/auth.ts', 5);
|
|
998
|
+
|
|
999
|
+
// v5.0.1 — Zero-config auto-setup
|
|
1000
|
+
const setup = new AutoSetup('./my-project');
|
|
1001
|
+
const config = await setup.detect();
|
|
1002
|
+
// → { projectType: 'fullstack', languages: ['typescript','python'], frameworks: ['next.js'], aiTools: [...] }
|
|
1003
|
+
await setup.installMCPForTools(['Claude Code', 'Cursor']);
|
|
1004
|
+
await setup.installGitHooks();
|
|
1005
|
+
|
|
1006
|
+
// Create the brain
|
|
1007
|
+
const brain = new Orchestrator({
|
|
1008
|
+
provider: 'anthropic',
|
|
1009
|
+
projectDir: './my-project',
|
|
1010
|
+
agents: ['claude-code', 'cursor', 'cline'],
|
|
1011
|
+
watchMode: true,
|
|
1012
|
+
autoInject: true,
|
|
1013
|
+
brainPersonality: 'balanced',
|
|
1014
|
+
});
|
|
1015
|
+
|
|
1016
|
+
// v5.0.1 — Plugin system
|
|
1017
|
+
const plugins = new PluginSystem('./my-project');
|
|
1018
|
+
await plugins.loadAll();
|
|
1019
|
+
await plugins.preAnalysis({ files: ['src/auth.ts'], projectDir: './my-project' });
|
|
1020
|
+
|
|
1021
|
+
// v5.0.1 — Brain portability
|
|
1022
|
+
const porter = new BrainPortability('./my-project');
|
|
1023
|
+
const { filePath, sizeBytes, checksum } = await porter.exportBrain();
|
|
1024
|
+
|
|
1025
|
+
// v5.0.0 — Infinite Intelligence
|
|
1026
|
+
const hmem = new HierarchicalMemory();
|
|
1027
|
+
await hmem.store('JWT rotation pattern: rotate every 24h with refresh tokens', 'security', 0.9);
|
|
1028
|
+
const results = hmem.search('jwt');
|
|
1029
|
+
|
|
1030
|
+
const recall = new ContextRecall(hmem);
|
|
1031
|
+
const activated = recall.recall({
|
|
1032
|
+
currentFile: 'src/auth/login.ts',
|
|
1033
|
+
currentCategory: 'security',
|
|
1034
|
+
recentEdits: ['src/auth/login.ts', 'src/middleware/auth.ts'],
|
|
1035
|
+
projectType: 'node',
|
|
1036
|
+
keywords: ['jwt', 'authentication', 'session'],
|
|
1037
|
+
});
|
|
1038
|
+
|
|
1039
|
+
const consensus = new ConsensusEngine('my-agent');
|
|
1040
|
+
const proposalId = consensus.propose('Use bcrypt cost factor 12', 'security', 0.95);
|
|
1041
|
+
|
|
1042
|
+
// Start everything
|
|
1043
|
+
await brain.start();
|
|
1044
|
+
```
|
|
1045
|
+
|
|
1046
|
+
---
|
|
1047
|
+
|
|
1048
|
+
## Health Score System
|
|
1049
|
+
|
|
1050
|
+
| Dimension | Weight | Checks |
|
|
1051
|
+
|---|---|---|
|
|
1052
|
+
| Security | 25% | Vulnerabilities, secrets, exposed keys |
|
|
1053
|
+
| Quality | 25% | Code smells, anti-patterns, complexity |
|
|
1054
|
+
| Performance | 15% | N+1 queries, memory leaks, blocking ops |
|
|
1055
|
+
| Architecture | 15% | Coupling, cohesion, dependencies |
|
|
1056
|
+
| Maintainability | 10% | Duplication, file sizes, comments |
|
|
1057
|
+
| Test Coverage | 10% | Test presence, assertion quality |
|
|
1058
|
+
|
|
1059
|
+
---
|
|
1060
|
+
|
|
1061
|
+
## Version History
|
|
1062
|
+
|
|
1063
|
+
> **Full changelogs, migration guides, and detailed release notes:** [`docs/versions/`](docs/versions/)
|
|
1064
|
+
>
|
|
1065
|
+
> **Compact changelog:** [`CHANGELOG.md`](CHANGELOG.md)
|
|
1066
|
+
|
|
1067
|
+
| Version | Codename | Date | Highlights |
|
|
1068
|
+
|---|---|---|---|
|
|
1069
|
+
| **[v5.1.1](docs/versions/v5.1.1.md)** | Hyper-Cognitive | 2026-04-18 | LSP Server, Fine-Tuning Engine, Smart Cache, Intent Engine, Code DNA, Temporal Intelligence |
|
|
1070
|
+
| **[v5.0.1](docs/versions/v5.0.1.md)** | Zero-Config | 2026-04-15 | Auto-Setup, MCP Server (19 tools), Dashboard, NLP Queries, Plugin System |
|
|
1071
|
+
| **[v5.0.0](docs/versions/v5.0.0.md)** | Infinite Intelligence | 2026-04-12 | Hierarchical Memory, Associative Recall, Multi-Agent Consensus, Collective Learning |
|
|
1072
|
+
| **[v4.0.0](docs/versions/v4.0.0.md)** | Hyper-Intelligence | 2026-04-08 | TurboQuant, SSSP Routing, Self-Evolution, CAIP, Adversarial Defense, Swarm, Knowledge Graph |
|
|
1073
|
+
| **[v3.0.0](docs/versions/v3.0.0.md)** | Hyper-Intelligence | 2026-04-01 | Cognitive Load Analysis, Security Audit Engine, Influence Map |
|
|
1074
|
+
| **[v2.1.0](docs/versions/v2.1.0.md)** | Quantum Neural Mesh | 2026-03-25 | Shannon Entropy, Cosine Similarity, Bayesian Confidence |
|
|
1075
|
+
| **[v2.0.0](docs/versions/v2.0.0.md)** | Super-Intelligence | 2026-03-18 | Semantic Analyzer, Dependency Graph, Code Similarity, MCP Server |
|
|
1076
|
+
| **[v1.0.0](docs/versions/v1.0.0.md)** | Foundation | 2026-02-28 | 7 Agent Adapters, Health Scoring, Smart Fixes, Vulnerability Scanning |
|
|
1077
|
+
|
|
1078
|
+
---
|
|
1079
|
+
|
|
1080
|
+
## Supported Agents & Tools
|
|
1081
|
+
|
|
1082
|
+
| Agent | Status | MCP Support | Injection Target |
|
|
1083
|
+
|---|---|---|---|
|
|
1084
|
+
| [Claude Code](https://docs.anthropic.com/en/docs/claude-code) | Full Support | Auto-install | `CLAUDE.md`, `.claude/` |
|
|
1085
|
+
| [Cursor](https://cursor.sh/) | Full Support | Auto-install | `.cursor/mcp.json` |
|
|
1086
|
+
| [Kilo Code](https://kilocode.ai/) | Full Support | Auto-install | `.kilocode/rules.md` |
|
|
1087
|
+
| [Cline](https://github.com/cline/cline) | Full Support | Auto-install | `.clinerules` |
|
|
1088
|
+
| [OpenCode](https://github.com/opencode-ai/opencode) | Full Support | Yes | `AGENT.md` |
|
|
1089
|
+
| [Codex](https://github.com/openai/codex) | Full Support | Yes | `AGENTS.md` |
|
|
1090
|
+
| [Roo Code](https://roocode.com/) | Full Support | Yes | `.roo/rules.md` |
|
|
1091
|
+
| [Aider](https://aider.chat/) | Full Support | Yes | `.aider.conf.yml` |
|
|
1092
|
+
| [Windsurf](https://codeium.com/windsurf) | Detection | Yes | `.windsurfrules` |
|
|
1093
|
+
|
|
1094
|
+
---
|
|
1095
|
+
|
|
1096
|
+
## Development
|
|
1097
|
+
|
|
1098
|
+
```bash
|
|
1099
|
+
git clone https://github.com/theihtisham/agent-shadow-brain.git
|
|
1100
|
+
cd agent-shadow-brain
|
|
1101
|
+
npm install
|
|
1102
|
+
npm run build
|
|
1103
|
+
npm test
|
|
1104
|
+
```
|
|
1105
|
+
|
|
1106
|
+
---
|
|
1107
|
+
|
|
1108
|
+
## Roadmap
|
|
1109
|
+
|
|
1110
|
+
- [x] ~~MCP Server, multi-project, team mode~~ (v2.0)
|
|
1111
|
+
- [x] ~~Cross-session neural mesh~~ (v2.1)
|
|
1112
|
+
- [x] ~~TurboQuant infinite memory~~ (v4.0)
|
|
1113
|
+
- [x] ~~SSSP quantum routing~~ (v4.0)
|
|
1114
|
+
- [x] ~~Self-evolving rules~~ (v4.0)
|
|
1115
|
+
- [x] ~~Cross-agent protocol~~ (v4.0)
|
|
1116
|
+
- [x] ~~Adversarial hallucination defense~~ (v4.0)
|
|
1117
|
+
- [x] ~~Hierarchical memory compression~~ (v5.0)
|
|
1118
|
+
- [x] ~~Context-triggered associative recall~~ (v5.0)
|
|
1119
|
+
- [x] ~~Multi-agent consensus protocol~~ (v5.0)
|
|
1120
|
+
- [x] ~~Collective cross-project learning~~ (v5.0)
|
|
1121
|
+
- [x] ~~Zero-config auto-setup~~ (v5.0.1)
|
|
1122
|
+
- [x] ~~MCP server for Cursor + all tools~~ (v5.0.1)
|
|
1123
|
+
- [x] ~~Rich web dashboard with real-time WebSocket~~ (v5.0.1)
|
|
1124
|
+
- [x] ~~Plugin system + brain export/import~~ (v5.0.1)
|
|
1125
|
+
- [x] ~~Custom LLM fine-tuning~~ (v5.1.1)
|
|
1126
|
+
- [x] ~~Language server (LSP)~~ (v5.1.1)
|
|
1127
|
+
- [x] ~~Web dashboard v3~~ (v5.1.1)
|
|
1128
|
+
- [ ] **Visual Studio Code integration** — inline insight overlays
|
|
1129
|
+
- [ ] **Multi-model orchestration** — route insights through multiple LLMs
|
|
1130
|
+
- [ ] **Real-time collaboration** — team brain with live cursors
|
|
1131
|
+
|
|
1132
|
+
---
|
|
1133
|
+
|
|
1134
|
+
## License
|
|
1135
|
+
|
|
1136
|
+
MIT License — see [LICENSE](LICENSE) for details.
|
|
1137
|
+
|
|
1138
|
+
---
|
|
1139
|
+
|
|
1140
|
+
## Author
|
|
1141
|
+
|
|
1142
|
+
**theihtisham**
|
|
1143
|
+
|
|
1144
|
+
[](https://github.com/theihtisham)
|
|
1145
|
+
[](https://www.npmjs.com/~theihtisham)
|
|
1146
|
+
|
|
1147
|
+
---
|
|
1148
|
+
|
|
1149
|
+
<div align="center">
|
|
1150
|
+
|
|
1151
|
+
### If Shadow Brain makes your AI agents smarter, give it a star
|
|
1152
|
+
|
|
1153
|
+
[](https://star-history.com/#theihtisham/agent-shadow-brain&Date)
|
|
1154
|
+
|
|
1155
|
+
**Built with brains by [theihtisham](https://github.com/theihtisham)**
|
|
1156
|
+
|
|
1157
|
+
**Topics:** `ai` `artificial-intelligence` `llm` `claude-code` `cursor` `cline` `kilo-code` `codex` `roo-code` `aider` `opencode` `windsurf` `code-review` `static-analysis` `developer-tools` `npm-package` `typescript` `ollama` `openai` `anthropic` `code-quality` `security-scanner` `health-score` `smart-fix` `vulnerability-scanner` `pattern-learning` `agent-tools` `ai-agent` `code-metrics` `pre-commit-hook` `github-actions` `developer-productivity` `neural-mesh` `quantum-computing` `shannon-entropy` `bayesian-inference` `mcp-server` `model-context-protocol` `semantic-analysis` `dependency-graph` `turboquant` `sssp-algorithm` `genetic-algorithm` `swarm-intelligence` `pagerank` `knowledge-graph` `hallucination-detection` `cross-agent` `self-evolving` `predictive-engine` `ant-colony` `infinite-memory` `vector-compression` `iclr-2026` `arxiv` `super-intelligence` `hyper-intelligence` `autonomous-agents` `ai-engineering` `agentic-coding` `vibe-coding` `cursor-ai` `zero-config` `auto-setup` `dashboard` `plugin-system` `brain-export` `natural-language` `real-time` `websocket` `team-collaboration` `ci-cd` `open-source` `trending` `hierarchical-memory` `context-recall` `consensus-engine` `collective-learning` `infinite-intelligence` `coding-assistant` `lsp-server` `language-server-protocol` `fine-tuning` `smart-cache` `intent-engine` `code-dna` `temporal-intelligence` `hyper-cognitive`
|
|
1158
|
+
|
|
1159
|
+
</div>
|