File size: 2,545 Bytes
6132d31
 
5cc86a1
 
 
6132d31
 
 
5cc86a1
6132d31
5cc86a1
 
a607ce7
6132d31
a607ce7
5cc86a1
6132d31
a607ce7
 
5cc86a1
 
a607ce7
 
5cc86a1
a607ce7
5cc86a1
a607ce7
 
5cc86a1
a607ce7
 
 
 
5cc86a1
 
a607ce7
 
 
5cc86a1
a607ce7
 
 
5cc86a1
a607ce7
 
 
5cc86a1
a607ce7
 
 
5cc86a1
a607ce7
 
 
5cc86a1
a607ce7
 
 
5cc86a1
a607ce7
 
 
5cc86a1
a607ce7
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
---
title: START HERE  Agent Forensics Suite
emoji: 🔥
colorFrom: indigo
colorTo: purple
sdk: gradio
sdk_version: 6.3.0
app_file: app.py
pinned: true
short_description: 'Start here: verify agent runs with receipts, replays, diffs.'
license: other
thumbnail: >-
  https://cdn-uploads.huggingface.co/production/uploads/
---

# RFTSystems — Agent Forensics Suite (START HERE)

Evidence-first instrumentation for AI agents and safety decisions.  
Audit, prove, replay, and diff runs — turning “trust me” into verification.

## Why this exists
AI is being deployed faster than teams can reproduce failures, attribute responsibility, or prove what changed.  
The unacceptable part isn’t that systems fail — it’s that **nobody can prove what happened**.

This suite exists to make runs **inspectable and independently verifiable**.

## The workflow
**learn → generate proof → record reality → seal it → replay → diff → audit → benchmark**

## Quick start (60 seconds)
1. Open **AuditPlane** and generate a baseline suite (signed receipts + Merkle proofs).
2. Replay the same suite and confirm **drift diffs**.
3. Export the offline bundle so anyone can verify it without trusting you.

## The labs
- **AuditPlane — LLM Decision Proofs**  
  Ed25519-signed receipts + hash-chained runs + replay + drift diffs + Merkle proofs  
  https://huggingface.co/spaces/RFTSystems/AuditPlane__LLM_Decision_Proofs

- **ReplayProof — Agent POV Verified Replay**  
  Deterministic run bundles you can replay and verify anywhere  
  https://huggingface.co/spaces/RFTSystems/ReplayProof__Agent_POV__Verified_Replay

- **Agent Flight Recorder**  
  Hash-chained event logging across prompts, tools, outputs, and memory reads/writes  
  https://huggingface.co/spaces/RFTSystems/Agent_Flight_Recorder

- **RFT Memory Receipt Engine**  
  Generate and verify tamper-evident receipts for agent state/memory  
  https://huggingface.co/spaces/RFTSystems/RFT_Memory_Receipt_Engine

- **TimelineDiff — Differential Reproducibility**  
  Align two bundles and pinpoint the first divergence  
  https://huggingface.co/spaces/RFTSystems/TimelineDiff__Differential_Reproducibility

- **TrustStack Console**  
  Audit cockpit: inspect runs, compare state, trace what changed and why  
  https://huggingface.co/spaces/RFTSystems/TrustStack_Console

- **Coherent Compute Engine**  
  Verification-first benchmark with downloadable receipts  
  https://huggingface.co/spaces/RFTSystems/Coherent_Compute_Engine

## Principle
No receipts → no claims.