rust-kgdb 0.5.4 → 0.5.6
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/CHANGELOG.md +49 -0
- package/README.md +643 -0
- package/examples/fraud-detection-agent.js +134 -0
- package/examples/hypermind-complete-demo.js +926 -0
- package/examples/underwriting-agent.js +127 -0
- package/package.json +1 -1
package/CHANGELOG.md
CHANGED
|
@@ -2,6 +2,55 @@
|
|
|
2
2
|
|
|
3
3
|
All notable changes to the rust-kgdb TypeScript SDK will be documented in this file.
|
|
4
4
|
|
|
5
|
+
## [0.5.6] - 2025-12-15
|
|
6
|
+
|
|
7
|
+
### Enhanced HyperMind Examples - Show REAL Power
|
|
8
|
+
|
|
9
|
+
This release dramatically improves the HyperMind framework examples to demonstrate the true value of neuro-symbolic AI.
|
|
10
|
+
|
|
11
|
+
#### New: Comprehensive Demo (`examples/hypermind-complete-demo.js`)
|
|
12
|
+
|
|
13
|
+
7-section demonstration showing the FULL HyperMind pipeline:
|
|
14
|
+
|
|
15
|
+
1. **Thought-Provoking Problem Statement** - Why current AI agents fail in regulated industries
|
|
16
|
+
2. **Data Justification** - Why we use insurance fraud data instead of LUBM benchmark
|
|
17
|
+
3. **Knowledge Graph Setup** - Real NICB/FBI fraud patterns loaded as RDF triples
|
|
18
|
+
4. **Embedding Pipeline** - Claims vectorized for semantic similarity (HNSW indexing)
|
|
19
|
+
5. **GraphFrame Exploration** - PageRank, triangles, connected components for fraud ring detection
|
|
20
|
+
6. **Datalog Reasoning** - NICB rules encoded for deterministic inference
|
|
21
|
+
7. **Agent Interaction** - Natural language prompts with auditable responses
|
|
22
|
+
8. **Value Comparison** - Clear table showing HyperMind vs Vanilla LLM vs DSPy
|
|
23
|
+
9. **WASM Sandbox Security** - Capability-based security model explained
|
|
24
|
+
|
|
25
|
+
#### Updated: Fraud Detection Agent (`examples/fraud-detection-agent.js`)
|
|
26
|
+
|
|
27
|
+
- Added **Phase 6: Conversational Agent Demonstration**
|
|
28
|
+
- Shows user prompt → agent reasoning → structured response
|
|
29
|
+
- Displays full Datalog rule derivation with matching facts
|
|
30
|
+
- Includes "WHY THIS MATTERS" value proposition section
|
|
31
|
+
- Evidence chains and cryptographic proof hashes
|
|
32
|
+
|
|
33
|
+
#### Updated: Underwriting Agent (`examples/underwriting-agent.js`)
|
|
34
|
+
|
|
35
|
+
- Added **Phase 6: Conversational Agent Demonstration**
|
|
36
|
+
- Shows underwriting queue prioritization conversation
|
|
37
|
+
- "Why is TransCo flagged for review?" follow-up demonstration
|
|
38
|
+
- Full rule firing display with matched facts
|
|
39
|
+
|
|
40
|
+
#### Updated: README.md
|
|
41
|
+
|
|
42
|
+
- Added **"HyperMind in Action"** section showing real agent output
|
|
43
|
+
- Complete conversation example with rule derivations
|
|
44
|
+
- Value comparison table (HyperMind vs Vanilla LLM)
|
|
45
|
+
- Execution witness JSON for audit trail
|
|
46
|
+
|
|
47
|
+
### Technical Details
|
|
48
|
+
|
|
49
|
+
- All examples tested and verified working
|
|
50
|
+
- Total demo runtime: ~300ms
|
|
51
|
+
- Real insurance fraud data based on NICB/FBI statistics
|
|
52
|
+
- Full type theory (Hindley-Milner) + category theory + proof theory demonstrated
|
|
53
|
+
|
|
5
54
|
## [0.3.0] - 2025-12-11
|
|
6
55
|
|
|
7
56
|
### Major New Features
|