Add clean and data-driven dossier editions

This commit is contained in:
root 2025-12-22 02:05:03 +00:00
parent 0e77226d29
commit 8c1c448f28
3 changed files with 417 additions and 13 deletions

View file

@ -31,9 +31,9 @@ This dossier documents the **InfraFabric microlab**: a functioning single-shard
| Core claim | Proof (artifacts) | Limitation (scope / boundary) |
|---|---|---|
| **A) Traceability is safety.** Highstakes agents cannot be trusted without a verifiable history of what happened (request → retrieval → decision → output). | **IF.TTT + evidence bundles**: the IF.emotion trace protocol ships a portable tarball + manifest + verifier steps that a third party can run.<br/>Start here: [IF.emotion trace protocol (v3.3, styled) — endtoend verification appendix](https://infrafabric.io/static/hosted/IF_EMOTION_DEBUGGING_TRACE_WHITEPAPER_v3.3_STYLED.md). | **Microlab / single shard.** Proven in a single-host environment. Completeness is bounded by explicit witness boundaries; PQ is anchored at registry time (not necessarily on every hot-path artifact). No public appendonly transparency log yet. |
| **B) Governance requires plurality.** A single model acting as “the judge” is brittle; adversarial viewpoints and escalation are required. | **IF.BIAS → IF.GUARD**: risk preflight sizes councils and escalates; councils preserve dissent and veto paths; decisions are traced. Pointers: IF.BIAS, IF.GUARD, IF.5W sections. | **Cost / latency tradeoffs.** Multi-seat governance is reserved for higher-stakes decisions; low-stakes paths use smaller councils or fast-track gates. |
| **C) Context is the best firewall.** Static filters fail; security must distinguish “reference” vs “leak” and “discussion” vs “exfiltration”. | **IF.ARMOUR + IF.YOLOGUARD**: epistemic/anomaly detection primitives and secret/relationship screening patterns (architecture + docs). | **Domain specificity.** Calibrated for concrete security surfaces (secrets/PII/prompt injection); generalizing to broader “harmful intent” is an open research vector. |
| **A) Traceability is safety.** Highstakes agents cannot be trusted without a verifiable history of what happened (request → retrieval → decision → output). | **IF.TTT + evidence bundle + verifier**<br/>- Paper: [IF.emotion trace protocol (v3.3, styled)](https://infrafabric.io/static/hosted/IF_EMOTION_DEBUGGING_TRACE_WHITEPAPER_v3.3_STYLED.md)<br/>- Verifier: `https://infrafabric.io/static/hosted/iftrace.py`<br/>- Reference bundle: `emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz` | **Microlab / single shard.** Proven in a single-host environment. Completeness is bounded by explicit witness boundaries; PQ is anchored at registry time (not necessarily on every hot-path artifact). No public appendonly transparency log yet. |
| **B) Governance requires plurality.** A single model acting as “the judge” is brittle; adversarial viewpoints and escalation are required. | **IF.BIAS → IF.GUARD**<br/>- IF.BIAS: risk preflight sizes councils and escalates<br/>- IF.GUARD: councils preserve dissent and veto paths<br/>- IF.5W: structured inquiry briefs for councils | **Cost / latency tradeoffs.** Multi-seat governance is reserved for higher-stakes decisions; low-stakes paths use smaller councils or fast-track gates. |
| **C) Context is the best firewall.** Static filters fail; security must distinguish “reference” vs “leak” and “discussion” vs “exfiltration”. | **IF.ARMOUR + IF.YOLOGUARD**<br/>- IF.ARMOUR: epistemic coherence checks (detective layer)<br/>- IF.YOLOGUARD: secret/relationship screening primitives | **Domain specificity.** Calibrated for concrete security surfaces (secrets/PII/prompt injection); generalizing to broader “harmful intent” is an open research vector. |
### Rosetta Stone (Closest Analog, not “equals”)
@ -322,13 +322,13 @@ This is the connective tissue for the corpus: each paper points to the next laye
```mermaid
flowchart TD
MASTER["Master Whitepaper\nINFRAFABRIC_MASTER_WHITEPAPER"] --> TTT["IF_TTT_THE_SKELETON_OF_EVERYTHING"]
MASTER["Master Whitepaper<br/>INFRAFABRIC_MASTER_WHITEPAPER"] --> TTT["IF_TTT_THE_SKELETON_OF_EVERYTHING"]
MASTER --> GUARD["IF_GUARD_COUNCIL_FRAMEWORK"]
MASTER --> PACKET["IF_PACKET_TRANSPORT_FRAMEWORK"]
GUARD --> FIVEW["IF_5W_STRUCTURED_INQUIRY_FRAMEWORK"]
TTT --> STORY["IF_STORY_NARRATIVE_LOGGING"]
TTT --> EMOTION["IF_EMOTION_WHITEPAPER"]
EMOTION --> EMOOPS["emo-social runtime\n(trace_log + RAG)"]
EMOTION --> EMOOPS["emo-social runtime<br/>(trace_log + RAG)"]
PACKET --> SWARM["IF_SWARM-S2-COMMS"]
```
@ -379,8 +379,7 @@ Next steps (TTT hardening): enforce “cite only retrieved chunks” in response
_Source: `Danny Stocker - CV - InfraFabric.pdf`_
**Contact**
- Email: danny.stocker@gmail.com
- Phone: +(33) 6 52 48 90 17
- Email: ds@infrafabric.io
- Web: https://digital-lab.ca/dannystocker
**Headline**
@ -439,6 +438,10 @@ InfraFabric is the operating system that turns AI from a chatbot into a reliable
## Index
- Submission editions (recommended first read):
- [Submission Edition (Clean)](DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION.md)
- [DataDriven Technical Report (Microlab)](DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION.md)
- [00. The Bridge: Submission Pack (Reviewer Orientation)](#bridge-submission-pack)
### External audit artifacts (public, reviewer-friendly)
@ -446,6 +449,7 @@ InfraFabric is the operating system that turns AI from a chatbot into a reliable
These artifacts are published in a dedicated repo and mirrored to a static directory for reliable downloads (avoids intermittent Forgejo “raw” quirks).
- Public static mirror (preferred): `https://infrafabric.io/static/hosted/`
- Dossier viewer (renders markdown + anchors): `https://infrafabric.io/static/hosted/dossier_site/index.html`
- Source repo: `https://git.infrafabric.io/danny/hosted`
Key artifacts:
@ -3445,7 +3449,7 @@ Special thanks to:
**License:** Creative Commons Attribution 4.0 International (CC BY 4.0)
**Code:** Available at https://git.infrafabric.io/dannystocker
**Contact:** InfraFabric Project (danny.stocker@gmail.com)
**Contact:** InfraFabric Project (ds@infrafabric.io)
---
@ -6854,7 +6858,7 @@ The InfraFabric project is open research—all methodologies, frameworks, and va
**License:** Creative Commons Attribution 4.0 International (CC BY 4.0)
**Code & Data:** Available at https://git.infrafabric.io/dannystocker
**Contact:** Danny Stocker (danny.stocker@gmail.com)
**Contact:** Danny Stocker (ds@infrafabric.io)
**arXiv Category:** cs.AI, cs.SE, cs.HC
---
@ -31072,12 +31076,12 @@ IF.bus mirrors this architecture for AI agent coordination and financial service
```mermaid
flowchart TD
BUS["IF.bus motherboard v2.0"] --> CHIPS["Core chips\nIF.guard • IF.witness • IF.yologuard • IF.emotion"]
BUS --> LANES["Bus lanes\nDDS • Redis pub/sub"]
BUS --> SLOTS["Expansion slots\nif.api adapters (9)"]
BUS["IF.bus motherboard v2.0"] --> CHIPS["Core chips<br/>IF.guard • IF.witness • IF.yologuard • IF.emotion"]
BUS --> LANES["Bus lanes<br/>DDS • Redis pub/sub"]
BUS --> SLOTS["Expansion slots<br/>if.api adapters (9)"]
BUS --> FIRMWARE["IF.ground firmware"]
BUS --> POWER["IF.connect power"]
SLOTS --> SLOT9["African fintech slot\n4 adapters"]
SLOTS --> SLOT9["African fintech slot<br/>4 adapters"]
CHIPS --> TTT["IF.TTT | Distributed Ledger traceability"]
```

View file

@ -0,0 +1,193 @@
# InfraFabric Dossier — DataDriven Technical Report (Microlab) v1.0
**Subject:** Measured characteristics of the IF.TTT trace pipeline (microlab)
**Protocol:** IF.TTT.dossier.metrics
**Status:** TECHNICAL REPORT (BORING ON PURPOSE)
**Date:** 2025-12-22
**Citation:** `if://doc/INFRAFABRIC_DOSSIER_DATA_DRIVEN/v1.0`
**Author:** Danny Stocker (`ds@infrafabric.io`)
**Web:** https://infrafabric.io
This edition intentionally avoids narrative framing. It reports what can be measured, what cannot, and what is planned.
**Canonical (static mirror):** `https://infrafabric.io/static/hosted/DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION.md`
**Repo source:** `https://git.infrafabric.io/danny/hosted/src/branch/main/DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION.md`
**SHA256 (sidecar):** `https://infrafabric.io/static/hosted/DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION.md.sha256`
**Verify:** `curl -fsSLO 'https://infrafabric.io/static/hosted/DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION.md' -fsSLO 'https://infrafabric.io/static/hosted/DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION.md.sha256' && sha256sum -c DANNY_STOCKER_INFRAFABRIC_DOSSIER_DATA_DRIVEN_EDITION.md.sha256`
---
## 1) Scope
This report covers:
- the IF.emotion evidence bundle format (tar.gz + manifest)
- measured latencies recorded in trace events (`auth_ms`, `rag_ms`, `llm_ms`) for a small sample of published bundles
- storage footprint of published bundles
This report does **not** claim:
- production scalability
- clinical validity
- “truth” of model outputs (only provenance of what the system did)
---
## 2) Data Sources (Public, Verifiable)
All artifacts referenced here are publicly downloadable and hash-verifiable.
Static mirror (preferred): `https://infrafabric.io/static/hosted/`
Source repo: `https://git.infrafabric.io/danny/hosted`
### 2.1 Evidence bundles used in this report
| Trace | Bundle | SHA256 sidecar |
|---|---|---|
| `016cca78-6f9d-4ffe-aec0-99792d383ca1` | `https://infrafabric.io/static/hosted/emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz` | `https://infrafabric.io/static/hosted/emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz.sha256` |
| `0642c357-7f8d-4eb5-9643-1992e7ee14a9` | `https://infrafabric.io/static/hosted/emo_trace_payload_0642c357-7f8d-4eb5-9643-1992e7ee14a9.tar.gz` | `https://infrafabric.io/static/hosted/emo_trace_payload_0642c357-7f8d-4eb5-9643-1992e7ee14a9.tar.gz.sha256` |
| `09aad3e1-f420-451e-a189-e86f68073dc0` | `https://infrafabric.io/static/hosted/emo_trace_payload_09aad3e1-f420-451e-a189-e86f68073dc0.tar.gz` | `https://infrafabric.io/static/hosted/emo_trace_payload_09aad3e1-f420-451e-a189-e86f68073dc0.tar.gz.sha256` |
| `96700e8e-6a83-445e-86f7-06905c500146` | `https://infrafabric.io/static/hosted/emo_trace_payload_96700e8e-6a83-445e-86f7-06905c500146.tar.gz` | `https://infrafabric.io/static/hosted/emo_trace_payload_96700e8e-6a83-445e-86f7-06905c500146.tar.gz.sha256` |
### 2.2 Verification command (bundle transport integrity)
```bash
curl -fsSLO '<BUNDLE_URL>' -fsSLO '<BUNDLE_URL>.sha256' && sha256sum -c '<BUNDLE_FILENAME>.sha256'
```
### 2.3 Verifier tool
- Static: `https://infrafabric.io/static/hosted/iftrace.py`
- Repo: `https://git.infrafabric.io/danny/hosted/raw/branch/main/iftrace.py`
Run:
```bash
python3 iftrace.py verify '<BUNDLE_FILENAME>.tar.gz'
```
---
## 3) Measurement Method (How Numbers Are Obtained)
For each bundle:
1. extract `payload/trace_events.jsonl`
2. read per-event `event.data`:
- `request_received.data.auth_ms`
- `retrieval_done.data.rag_ms` (when present)
- `model_done.data.llm_ms` (when present)
3. treat these values as **self-reported microlab timings** (they are not externally attested)
Key point: even if the numbers are not “audited”, the bundle makes them *replayable* and makes the presence/absence of events *auditable*.
---
## 4) Architecture Boundary (Where Guarantees Begin)
```mermaid
flowchart TB
U[User] -->|HTTPS| E[Edge]
E --> B[Backend Witness Boundary]
B --> R[Retrieval]
B --> P[Prompt]
B --> M[Model]
B --> X[Postprocess]
B --> T1["REQ_SEEN ledger<br/>(hourly JSONL)"]
B --> T2["Trace events<br/>(hash chain JSONL)"]
B --> T3["Signed summary<br/>(output hash + head attestation)"]
T1 --> H["Signed Merkle head<br/>(per hour)"]
T2 --> S["Trace head<br/>(event_hash)"]
H --> BUNDLE["Evidence bundle<br/>(tar.gz + manifest)"]
S --> BUNDLE
T3 --> BUNDLE
BUNDLE --> MIRROR["Static mirror<br/>(public download)"]
```
Interpretation: integrity begins at the backend witness boundary; completeness is meaningful at and after that boundary until edge witnessing exists.
---
## 5) Observed Sample Metrics (N=4 Bundles)
### 5.1 Bundle sizes (storage footprint)
| Trace | Outcome | Bundle size |
|---|---|---:|
| `016cca78-6f9d-4ffe-aec0-99792d383ca1` | full trace (retrieval + model) | 82,010 bytes |
| `0642c357-7f8d-4eb5-9643-1992e7ee14a9` | full trace (retrieval + model) | 5,515 bytes |
| `09aad3e1-f420-451e-a189-e86f68073dc0` | full trace (retrieval + model) | 71,817 bytes |
| `96700e8e-6a83-445e-86f7-06905c500146` | guard short-circuit (no retrieval/model) | 82,410 bytes |
Notes:
- N is small; treat these as indicative examples, not stable distributions.
- The short-circuit bundle being large indicates that “blocked paths” can still carry substantial evidence payloads (depending on included artifacts).
### 5.2 Latency fields recorded in trace events
| Trace | `auth_ms` | `rag_ms` | `llm_ms` | `retrieved_count` | Notes |
|---|---:|---:|---:|---:|---|
| `016cca78-6f9d-4ffe-aec0-99792d383ca1` | 3 | 1107 | 10550 | 1 | request → retrieval → model |
| `0642c357-7f8d-4eb5-9643-1992e7ee14a9` | 4 | 383 | 12287 | 2 | request → retrieval → model |
| `09aad3e1-f420-451e-a189-e86f68073dc0` | 4 | 1377 | 17879 | 2 | request → retrieval → model |
| `96700e8e-6a83-445e-86f7-06905c500146` | 5 | (n/a) | (n/a) | (n/a) | guard short-circuit reason: `self_harm_signal` |
Derived from the sample (successful traces only; N=3):
- `rag_ms`: min 383, median 1107, max 1377
- `llm_ms`: min 10550, median 12287, max 17879
---
## 6) What This Sample Demonstrates
From the published bundles, an external reviewer can verify (cryptographically / structurally):
- the chain-of-custody wiring exists (hash chain + signed summary + inclusion proof)
- “short-circuit” decisions are still witnessed and included in the trace log (a critical property for dispute resolution)
An external reviewer cannot verify (from the bundle alone):
- the correctness of the output in the outside world
- the completeness of events *before* the backend witness boundary (edge drops / load balancer denials)
- the integrity of local keys (key custody, rotation, compromise response)
---
## 7) Engineering Roadmap (Metrics-Driven)
This roadmap is written as measurable deliverables.
### 7.1 Key management hardening
- Add explicit key separation: `IF_REQ_SEEN_HMAC_KEY` must not fall back to signing secret.
- Document key rotation procedure and compromise response.
- Optional: HSM/TPM signing for production deployments.
### 7.2 Edge witnessing (completeness boundary expansion)
- Implement edge request attempt ledger (cryptographic, not just web logs).
- Publish a “completeness SLO” (e.g., signed head anchored every N minutes or every N requests).
### 7.3 Time and truncation defenses
- Add monotonic counters (per trace and per ledger hour).
- Anchor chain heads periodically to reduce tail truncation risk window.
### 7.4 Scale testing harness
- Replay harness that runs the verifier across 1000+ generated traces (including failures) and publishes summary metrics.
---
## 8) Companion Documents
- Full dossier (uncut): `DANNY_STOCKER_INFRAFABRIC_DOSSIER.md`
- IF.emotion trace protocol (detailed, with walkthrough): `https://infrafabric.io/static/hosted/IF_EMOTION_DEBUGGING_TRACE_WHITEPAPER_v3.3_STYLED.md`
- Evidence bundles directory: `https://infrafabric.io/static/hosted/`

View file

@ -0,0 +1,207 @@
# InfraFabric Dossier — Submission Edition (Clean) v1.0
**Subject:** Safety-first agent runtime with verifiable provenance (microlab)
**Protocol:** IF.TTT.dossier.submission
**Status:** SUBMISSION EDITION (CLEAN)
**Date:** 2025-12-22
**Citation:** `if://doc/INFRAFABRIC_DOSSIER_SUBMISSION/v1.0`
**Author:** Danny Stocker (`ds@infrafabric.io`)
**Web:** https://infrafabric.io
> The models answer is ephemeral. The trace is the product. If you cant prove what happened, you are not running an AI system — you are running a scripted reality show.
**Canonical (static mirror):** `https://infrafabric.io/static/hosted/DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION.md`
**Repo source:** `https://git.infrafabric.io/danny/hosted/src/branch/main/DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION.md`
**SHA256 (sidecar):** `https://infrafabric.io/static/hosted/DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION.md.sha256`
**Verify:** `curl -fsSLO 'https://infrafabric.io/static/hosted/DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION.md' -fsSLO 'https://infrafabric.io/static/hosted/DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION.md.sha256' && sha256sum -c DANNY_STOCKER_INFRAFABRIC_DOSSIER_SUBMISSION_EDITION.md.sha256`
**Read this if you have 15 minutes:** start at “Independent Verification (One Trace)”, run the commands, then skim “Boundaries & Limitations”.
**Read this if you have 60 minutes:** add “Architecture” + “Threat Model”.
**For full lab notes / uncut corpus:** see `DANNY_STOCKER_INFRAFABRIC_DOSSIER.md` (≈1.5MB).
---
## 1) Technical Disclosure (AI-Native Implementation)
This project follows an Operator-as-Architect workflow: I define constraints, interfaces, and verification hooks; LLMs generate implementation code under supervision. The system is therefore evaluated as:
- **Architecture + specifications:** authored by me (the claims in this submission reference verification paths).
- **Code implementation:** LLM-assisted and treated as *untrusted until audited*; the trace protocol is designed to make that audit possible.
---
## 2) What This Is (And What It Is Not)
**This is:** a working microlab proving that agentic systems can ship **portable, thirdparty verifiable evidence bundles** for disputed behaviors (request → retrieval → decision → output).
**This is not:** a claim that outputs are “true”, a claim of clinical efficacy, or a claim of production scale.
---
## 3) Core Claims, Proofs, Boundaries
| Core claim | Proof (artifact) | Boundary / limitation |
|---|---|---|
| **A) Traceability is a safety primitive.** Highstakes agent actions require verifiable provenance. | IF.emotion trace protocol + evidence bundle + verifier (see §5). | **Microlab / single shard.** Guarantees begin at the backend witness boundary; edge completeness is future work. |
| **B) Governance needs structured dissent.** One-model “judge” patterns are brittle. | IF.BIAS (risk preflight) + IF.GUARD (council with required dissent seat) traced via IF.TTT. | **Cost/latency.** Multi-seat governance is reserved for higher-stakes decisions; low-stakes paths stay fast. |
| **C) Context is a security control.** Static filters fail; provenance + coherence checks reduce unsafe drift. | IF.ARMOUR (coherence/detective layer) + IF.YOLOGUARD (secret/relationship screening) integrated into the pipeline. | **Domain calibration.** Strongest for concrete surfaces (secrets/PII/prompt injection); general harmful-intent is open research. |
---
## 4) Prior Art (Where This Fits)
InfraFabric is not “inventing audit logs”. It adapts known integrity patterns to a different object:
- **SLSA / SBOM / in-toto / Sigstore**: provenance for *software artifacts* → InfraFabric applies provenance discipline to *semantic decisions*, retrieval lineage, and agent actions.
- **Certificate Transparency (CT)**: append-only Merkle logging → InfraFabric borrows “inclusion proof” thinking for bounded completeness (`REQ_SEEN`).
- **Event sourcing / OpenTelemetry**: observability → InfraFabric adds a **verifier-oriented evidence bundle** that can be audited without trusting the operator.
- **W3C PROV (conceptually)**: provenance graphs → InfraFabric focuses on portable, hash-addressed artifacts with explicit boundaries and replay hooks.
The key differentiator is not “we log more”. It is: **we ship a dispute bundle that a hostile reviewer can verify.**
---
## 5) Independent Verification (One Trace, End-to-End)
This is the minimal “receipt” test: download one evidence bundle, verify transport integrity, then verify the internal chain-of-custody.
### 5.1 Public artifacts (preferred links)
Static mirror (preferred for reviewers): `https://infrafabric.io/static/hosted/`
Source repo (canonical): `https://git.infrafabric.io/danny/hosted`
Reference proof bundle:
- Bundle (static): `https://infrafabric.io/static/hosted/emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz`
- Bundle SHA256 (static): `https://infrafabric.io/static/hosted/emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz.sha256`
- Bundle (repo): `https://git.infrafabric.io/danny/hosted/raw/branch/main/emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz`
Verifier:
- `iftrace.py` (static): `https://infrafabric.io/static/hosted/iftrace.py`
- `iftrace.py` (repo): `https://git.infrafabric.io/danny/hosted/raw/branch/main/iftrace.py`
### 5.2 Verify transport integrity (hash)
```bash
curl -fsSLO 'https://infrafabric.io/static/hosted/emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz' \
-fsSLO 'https://infrafabric.io/static/hosted/emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz.sha256'
sha256sum -c 'emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz.sha256'
```
### 5.3 Verify internal chain-of-custody
```bash
curl -fsSLO 'https://infrafabric.io/static/hosted/iftrace.py'
python3 iftrace.py verify 'emo_trace_payload_016cca78-6f9d-4ffe-aec0-99792d383ca1.tar.gz'
```
The verifier checks:
- `trace_events.jsonl` forms a hash chain (`prev_hash``event_hash`)
- `req_seen_inclusion_proof.json` verifies against the signed Merkle head (bounded completeness for that hour)
- `ttt_signed_record.json` binds output hash + chain head and verifies signature
- the bundle manifest hashes match the extracted artifacts
### 5.4 What you should see
You should be able to verify, from raw artifacts inside the tarball, at minimum:
- `request_received` (auth/provider/model intent)
- `retrieval_done` (retrieval ids + `rag_ms`)
- `prompt_built` (prompt hash)
- `model_done` (`llm_ms`, model/provider)
- `trace_finalizing` (chain head)
- `response.json` (the actual assistant output bytes whose hash is bound in the signed summary)
If any element is missing or fails verification, the system is explicitly “not proven”.
---
## 6) Architecture (Where Guarantees Begin)
This diagram shows the system boundary that the protocol currently guarantees.
```mermaid
flowchart TB
U[User] -->|HTTPS| E[Edge]
E --> B[Backend Witness Boundary]
B --> R[Retrieval]
B --> P[Prompt]
B --> M[Model]
B --> X[Postprocess]
B --> T1["REQ_SEEN ledger<br/>(hourly JSONL)"]
B --> T2["Trace events<br/>(hash chain JSONL)"]
B --> T3["Signed summary<br/>(output hash + head attestation)"]
T1 --> H["Signed Merkle head<br/>(per hour)"]
T2 --> S["Trace head<br/>(event_hash)"]
H --> BUNDLE["Evidence bundle<br/>(tar.gz + manifest)"]
S --> BUNDLE
T3 --> BUNDLE
BUNDLE --> MIRROR["Static mirror<br/>(public download)"]
BUNDLE --> REG["Registry anchor<br/>(PQ-hybrid, when enabled)"]
```
Interpretation: integrity begins at the backend witness boundary; completeness is meaningful at and after that boundary until edge witnessing exists.
---
## 7) Threat Model (Why Standard Logs Fail)
The trace protocol is built for an explicit adversary: **the dispute**.
Primary failure modes it is designed to prevent:
- “The logs say one thing, the user screenshot says another.” (repudiation)
- “A retrieval set was poisoned, then the retrieval event was deleted.” (selective deletion)
- “A decision was claimed to be reviewed, but there is no lineage.” (audit theater)
Failure mode analysis (separate artifact):
- `https://infrafabric.io/static/hosted/IF_TTT_FAILURE_MODE_ANALYSIS_v1.md`
---
## 8) Validation Status (Honest)
- **External human testing:** anecdotal pre-testing only; not clinical validation.
- **Scale:** unproven beyond microlab; numbers are not presented as global claims.
- **Key management / attestations:** not audit-grade yet (explicit roadmap below).
---
## 9) Boundaries & Roadmap (What Must Be Built Next)
This is the “reviewer redflags” list, answered directly.
### 9.1 Key management (blocker for certification)
Current: file-based Ed25519 keys sufficient for microlab evidence binding.
Required: key generation ceremony + rotation + compromise response; HSM/TPM-backed signing for high-stakes deployments.
### 9.2 Completeness boundary (edge witnessing)
Current: bounded completeness begins at backend witness boundary.
Required: edge-level request witnessing (or independent transparency log) to reduce “selective trace” risk.
### 9.3 Code identity / execution integrity
Current: bundle binds outputs and retrieval lineage.
Required: signed deploy attestations; reproducible build digest binding; optional remote attestation in higher-assurance environments.
### 9.4 Clock integrity
Current: timestamps are informative, not cryptographically trusted.
Required: monotonic counters + periodic anchoring; external timestamping strategy for forensic-grade timelines.
---
## 10) Where to Go Deeper
- Full dossier (uncut): `DANNY_STOCKER_INFRAFABRIC_DOSSIER.md`
- IF.emotion trace protocol (detailed): `https://infrafabric.io/static/hosted/IF_EMOTION_DEBUGGING_TRACE_WHITEPAPER_v3.3_STYLED.md`
- Evidence bundles directory: `https://infrafabric.io/static/hosted/`