Dina — Test Results

The Architecture of Agency
Generated 2026-03-23 11:26:10 · 10m51s total runtime
97.4%
Overall Pass Rate
4970
Total
4841
Passed
31
Failed
34
XFail
SuiteTotalPass FailSkipXFail Status
Core (Go)201719830340PASS
Brain (Py)7467160129PASS
CLI (Py)109109000PASS
Admin CLI (Py)5959000PASS
AppView (TS)295295000PASS
Integration983978050PASS
E2E (Docker)121109075PASS
Release1391220170PASS
Install7265700FAIL
User Stories106822400FAIL
AppView Integration (TS)323323000PASS
TOTAL49704841 316434 FAILED

DINA User Story Tests

Stack: 2x Go Core + 2x Python Brain + PDS + AppView + Postgres — zero mocks, real crypto, real trust
77.4%
106 tests · 82 passed · 24 failed
01The Purchase Journey8/13 passed

“I need a chair” → 5 reviewers created (3 verified Ring 2, 2 unverified Ring 1)

Dina checks health vault (back pain, needs lumbar), finance vault (budget 10–20K INR)

Trust-weighted reviews: skip CheapChair (low trust score), recommends ErgoMax Elite

02The Sancho Moment2/7 passed

Sancho arrives → Sancho’s Dina contacts your Dina (D2D encrypted, Ed25519 signed)

Your Dina searches vault by Sancho’s DID, finds: “his mother had a fall”, “likes cardamom tea”

Nudge: “Sancho 15 min away. Ask about his sick mother. Make cardamom tea.”

03The Dead Internet Filter8/8 passed

“Is this video AI?” → Dina resolves creator DID via AT Protocol Trust Network

Elena (Ring 3): 200 attestations, 15 peer vouches, 2yr history → “authentic, trusted creator”

BotFarm (Ring 1): 0 attestations, 3‑day‑old account → “unverified, check other sources”

04The Persona Wall5/11 passed

Shopping agent asks “any health conditions?” → Guardian blocks cross-persona access

Health (restricted): “L4‑L5 herniation” withheld. Proposes “chronic back pain” only

User approves minimal disclosure. PII scrubber confirms no diagnosis leaked

05The Agent Gateway10/10 passed

OpenClaw/Perplexity Computer wants to send email → pairs with Home Node, asks Dina first

Dina checks: safe? matches your rules? PII leaking? “send_email” → MODERATE, asks you first

Safe tasks (web search) pass silently. Rogue agent with no auth → 401, blocked at the gate

06The License Renewal9/10 passed

User uploads license scan → Brain extracts fields with confidence scores

Deterministic reminder fires 30 days before expiry (no LLM in the scheduling)

Delegation: Brain generates strict JSON for DMV-Bot. Guardian flags for human review

07The Daily Briefing5/5 passed

Most noise waits quietly. Real harm interrupts immediately.

At the end of the day, Dina gives one calm summary and clears the queue.

08Move to a New Machine3/8 passed

Dina exports from the old machine and imports on the new one as an encrypted archive.

The wrong seed cannot unlock the vault. The same seed restores identity and data.

Migration is non‑destructive: the old machine still works after export.

09Connector Credential Expiry5/5 passed

Gmail OAuth expires — connector status: expired. Vault and identity still work.

User reconfigures credentials, connector resumes. No cascade, no crash.

10The Operator Journey5/5 passed

Re-run install script — DID unchanged (idempotent). No rotation, no orphaned data.

Identity is derived from master seed — immutable after bootstrap.

11The Anti-Her7/8 passed

“Haven’t talked to Sarah in 45 days” → proactive nudge in briefing, not on demand.

Life event follow-up: “Sancho’s mother was ill” → “you might want to check in.”

Emotional dependency detected → Dina suggests specific humans, never herself.

12Verified Truth8/9 passed

When Dina has little evidence, she says so honestly.

When people disagree, she says the evidence is mixed instead of pretending certainty.

When the signal is strong, she speaks clearly and points back to the original sources.

13Silence Under Stress3/3 passed

Even in a flood of alerts, Dina interrupts only for what truly matters.

Fake urgency from strangers is suspicious; trusted urgent events can break through.

14Agent Sandbox4/4 passed

No auth, no access. Revoked means revoked immediately.

Sensitive actions stay blocked unless you approve them.

Agents cannot impersonate someone else.

Core (Go)

98.3%
2017 tests · 1983 passed ·34 skipped
#Section TotalPass FailSkip Status
1Authentication & Authorization13513500PASS
2Key Derivation & Cryptography18318300PASS
3Identity (DID)11611600PASS
4Vault (SQLCipher)15315102PASS
5PII Scrubber (Tier 1 — Go Regex)353500PASS
6Gatekeeper (Egress / Sharing Policy)828200PASS
7Transport Layer878700PASS
8Task Queue (Outbox Pattern)454500PASS
9WebSocket Protocol444400PASS
10Device Pairing191900PASS
11Brain Client & Circuit Breaker252500PASS
12Admin Proxy7700PASS
13Rate Limiting6600PASS
14Configuration222002PASS
15API Endpoint Tests464600PASS
16Error Handling & Edge Cases101000PASS
17Security Hardening353104PASS
18Core ↔ Brain API Contract151500PASS
19Onboarding Sequence141400PASS
20Observability & Self-Healing333300PASS
21Logging Policy222200PASS
22PDS Integration (AT Protocol)212100PASS
23Portability & Migration282701PASS
24Deferred (Phase 2+)261025PASS
25Bot Interface4400PASS
26Client Sync Protocol7700PASS
27Digital Estate111100PASS
28CLI Request Signing (Ed25519)202000PASS
29Adversarial & Security (Behavioral)17217200PASS
30Test System Quality & Infrastructure32032000PASS
31Code Review Fix Verification545400PASS
32Security Fix Verification (Batch 5)131300PASS
33Additional Architecture-Review Coverage141400PASS
34Thesis Invariants — Loyalty10710700PASS
35Thesis Invariants — Silence First525200PASS
36Thesis Invariants — Action Integrity343400PASS
TOTAL 2017 1983 0 34

Brain (Py)

96.0%
746 tests · 716 passed ·1 skipped ·29 xfail
#Section TotalPass FailSkip Status
1Authentication & Authorization191900PASS
2Guardian Loop (Core AI Reasoning)15615600PASS
3PII Scrubber (Tier 2 — spaCy NER)696801PASS
4LLM Router (Multi-Provider)252500PASS
5Sync Engine (Ingestion Pipeline)939300PASS
6MCP Client (Agent Delegation)858500PASS
7Core Client (HTTP Client for dina-core)161600PASS
8Admin UI333300PASS
9Configuration101000PASS
10API Endpoints212100PASS
11Error Handling & Resilience9900PASS
12Scratchpad (Cognitive Checkpointing)121200PASS
13Crash Traceback Safety8800PASS
14Embedding Generation313100PASS
15Silence Classification Edge Cases6600PASS
16Anti-Her Enforcement5500PASS
17Thesis Invariants — Human Connection (Anti-Her)17800PASS
18Thesis Invariants — Silence First (Edge Cases)151100PASS
19Thesis Invariants — Pull Economy & Verified Truth20800PASS
20Thesis Invariants — Action Integrity (Brain-Side)12800PASS
21Deferred (Phase 2+)191900PASS
22Voice STT Integration3300PASS
23Code Review Fix Verification535300PASS
24Additional Architecture-Review Coverage9900PASS
TOTAL 746 716 0 1

CLI (Py)

100.0%
109 tests · 109 passed
#Section TotalPass FailSkip Status
1CLI (Py)10910900PASS
TOTAL 109 109 0 0

Admin CLI (Py)

100.0%
59 tests · 59 passed
#Section TotalPass FailSkip Status
1Admin CLI (Py)595900PASS
TOTAL 59 59 0 0

AppView (TS)

100.0%
295 tests · 295 passed
#Section TotalPass FailSkip Status
1Scorer Algorithms777700PASS
2Ingester Components838300PASS
3Shared Utilities383800PASS
4Configuration252500PASS
5API Cache (SWR)141400PASS
6Jetstream Consumer252500PASS
7Scorer Jobs171700PASS
8xRPC Params161600PASS
TOTAL 295 295 0 0

Integration

99.5%
983 tests · 978 passed ·5 skipped
#Section TotalPass FailSkip Status
1Core ↔ Brain Communication161600PASS
2End-to-End User Flows505000PASS
3Dina-to-Dina Communication434300PASS
4LLM Integration161600PASS
5Docker Networking & Isolation535300PASS
6Crash Recovery & Resilience131300PASS
7Security Boundary Tests898900PASS
8Digital Estate (SSS Custodian Recovery)141400PASS
9Ingestion-to-Vault Pipeline (Full E2E)484800PASS
10Data Flow Patterns (E2E)666600PASS
11Trust Network Integration494900PASS
12Upgrade & Migration141400PASS
14Chaos Engineering7700PASS
15Compliance & Privacy6600PASS
16Deferred (Phase 2+)757500PASS
17Architecture Validation (Cross-Cutting)151500PASS
18Architecture Validation — MEDIUM Severity Gaps848400PASS
19Thesis Invariants — Loyalty (Integration)10410400PASS
20Thesis Invariants — Human Connection (Integration)363600PASS
21Thesis Invariants — Silence First (Integration)818100PASS
22Thesis Invariants — Pull Economy & Verified Truth (Integration)565600PASS
23Thesis Invariants — Action Integrity (Integration)414100PASS
24Async Approval Flow7205PASS
TOTAL 983 978 0 5

E2E (Docker)

90.1%
121 tests · 109 passed ·7 skipped ·5 xfail
#Section TotalPass FailSkip Status
1Onboarding & First Run6600PASS
2Sancho Moment (Arrival Flow)6600PASS
3Product Research & Purchase6600PASS
4Memory & Recall4400PASS
5Ingestion Pipeline6600PASS
6Agent Safety & Delegation6600PASS
7Privacy & PII4400PASS
8Sensitive Personas4400PASS
9Digital Estate4400PASS
10Resilience & Recovery6600PASS
11Multi-Device Sync6600PASS
12Trust Network6600PASS
13Security & Adversarial101000PASS
14Agentic LLM Behavior4202PASS
15CLI Ed25519 Request Signing8800PASS
16AT Protocol PDS Integration7205PASS
17The Quiet Dina (Silence Protocol)3300PASS
18Move to a New Machine3300PASS
19Connector Failure & Recovery3300PASS
20Operator & Upgrade Journeys3300PASS
21Anti-Her (Thesis Invariant)5200PASS
22Verified Truth (Thesis Invariant)5300PASS
23Silence Stress (Thesis Invariant)3300PASS
24Agent Sandbox (Thesis Invariant)3300PASS
TOTAL 121 109 0 7

Release

87.8%
139 tests · 122 passed ·17 skipped
#Section TotalPass FailSkip Status
1Fresh Machine Install6600PASS
2First Conversation4400PASS
3Vault Persistence Across Restart4400PASS
4Locked-State and Seal Verification4400PASS
5Recovery Phrase & Disaster Recovery4400PASS
6Two Dinas Talk to Each Other4400PASS
7PDS and Trust Network E2E6600PASS
8Agent Gateway (Real/Rogue Client)4400PASS
9Persona Wall and PII Leakage4202PASS
10Hostile-Network D2D & Sancho Moment5500PASS
11Failure Handling & Degraded Operation8800PASS
12README & Public Claims Check9900PASS
15Install Re-Run (Idempotent)4400PASS
16Upgrade Verification3201PASS
17Admin Access Lifecycle4400PASS
18Connector Outage & Re-Auth4400PASS
19Silence Protocol & Daily Briefing4400PASS
20Draft-Don't-Send & Cart Handover4400PASS
21Export / Import Portability4301PASS
22External Exposure Audit5401PASS
23CLI Agent Integration & Pairing101000PASS
24Recommendation Integrity7106PASS
25Anti-Her & Staging Pipeline15906PASS
26Silence Stress7700PASS
27Action Integrity6600PASS
TOTAL 139 122 0 17

Install

90.3%
72 tests · 65 passed ·7 failed
#Section TotalPass FailSkip Status
1Install726570FAIL
TOTAL 72 65 7 0

AppView Integration (TS)

100.0%
323 tests · 323 passed
#Section TotalPass FailSkip Status
1Ingester Handlers616100PASS
2Deletion & Tombstones202000PASS
3Trust Edge Sync121200PASS
4Subject Resolution151500PASS
5Idempotency7700PASS
6Backpressure & Watermark101000PASS
7Rate Limiter5500PASS
8Graph Queries202000PASS
9Scorer Jobs444400PASS
10API Endpoints484800PASS
11Database Schema252500PASS
12Dirty Flags9900PASS
13Cursor Management6600PASS
14Backfill Script101000PASS
15Label Service6600PASS
16Docker Integration9900PASS
17End-to-End Flows111100PASS
18Web Server5500PASS
TOTAL 323 323 0 0