Trust Center

Trust is an inspectable surface, not a logo strip.

Mindburn keeps the public story tied to real artifacts: kernel source, Claim Matrix, proof docs, receipts, EvidencePacks, security posture, and data boundaries.

Verify HELM View HELM AI Kernel
Claims
Matrix-gated
Proof
Verifier-backed
Data
No public evidence rooms

Proof ledger

Real assets replace fake proof.

No invented logos, testimonials, pricing pages, unsupported KPIs, or private production proof. Each row has a source and review posture.

Surface Status Evidence Review Action
HELM AI Kernel repository Public Apache-2.0 execution kernel for validating model-proposed tool execution. PUBLIC Mindburn-Labs/helm-ai-kernel public repository 2026-05-12 stale risk Low Open repo →
HELM documentation Canonical public technical and product documentation. LIVE helm.docs.mindburn.org public documentation 2026-05-12 stale risk Low Open docs →
Source details Public wording mapped to source, status, and disclosure boundaries. PUBLIC src/data/claims.ts + /trust/claims 2026-05-12 stale risk Low Review details →
HELM verification route Receipt simulation, tamper check, source links, and unavailable console gates. PUBLIC mindburn/src/pages/trust/verification/index.astro 2026-05-21 stale risk Low Verify HELM →
Claim Matrix Public wording is gated as shipped, gated, target, research, or forbidden. PUBLIC helm-ai-enterprise/docs/CLAIM_MATRIX.md + mindburn /trust/claims 2026-05-24 stale risk Medium Review matrix →
Code Intelligence Graph proof Source-backed code graph evidence for engineering specs, affected tests, write scope, and closure evidence. GATED helm-ai-enterprise/docs/proof/code-intelligence-graph.md 2026-05-24 stale risk Medium Read spec gates →
MCP and tool-output quarantine Unknown MCP servers, hostile tool output, code comments, and README text are treated as untrusted evidence. GATED helm-ai-enterprise/docs/proof/mcp-quarantine-demo.md + tool-output-quarantine-demo.md 2026-05-24 stale risk Medium Review security →
Analog executor proof Analog actions require bounded scope, approvals, compensation, jurisdiction boundaries, receipts, and EvidencePacks. GATED helm-ai-enterprise/docs/proof/analog-executor-demo.md 2026-05-24 stale risk Medium Review proof →
Kinetic gateway proof Simulator-labeled AMR and factory gateway proof requires safety profile, telemetry, emergency halt posture, approvals, and EvidencePacks. SIMULATOR GATED helm-ai-enterprise/docs/proof/kinetic-gateway-demo.md 2026-05-24 stale risk High Review gateway →
Website SBOM Public site package inventory, not a third-party assurance report. PUBLIC public/security/sbom.cdx.json 2026-05-12 stale risk Medium Open SBOM →
Responsible disclosure Private vulnerability reporting path and data-submission boundaries. PUBLIC Mindburn public security page 2026-05-12 stale risk Low Read security →

Trust map

Six checks before the public story asks for belief.

Trust surfaces need visible boundaries, evidence, safety posture, data limits, and escalation paths.

01 Boundary

Action checks before effects

HELM describes authority at the action boundary, not as model confidence.

02 Data

Public channels stay bounded

Forms, email, and assistant routes are not private evidence rooms.

03 Evidence

Evidence maps to source

Public wording stays close to the source material that backs it.

04 Claim Matrix

Claims have statuses

Shipped, gated, simulator-gated, target, research, and forbidden wording stays visible.

05 Safety

Missing authority fails closed

Actions deny or escalate when policy, approval, scope, or proof needs are missing.

06 Escalation

Reports use dedicated routes

Security disclosure, architecture review, and public research paths stay separate.

Receipt verification demo

AI proposes an action. HELM decides whether it may run. The receipt makes the decision checkable later.

ALLOWT1 / dispatched
ProposeDecideReceipt
Send customer messageALLOW · customer_message.low_risk.v1 · rcpt-demo-d82c6ec7

Tamper detectable: changing the verdict produces a different receipt.

Status map

Every product surface has a status.

REVIEWED ACCESS

HELM AI Enterprise

Reviewed access / proof-backed

Canonical HELM AI Enterprise surface under UCS v1.3.
GATED

HELM AI Individual

Proof-gated / reviewed access

Canonical HELM AI Individual surface under UCS v1.3.
PUBLIC

HELM AI Kernel

Public

Canonical HELM AI Kernel surface under UCS v1.3.
GATED

CompanyArtifactGraph

Proof-gated

CompanyArtifactGraph is not execution authority.
GATED

Code Intelligence Graph

Proof-gated

Code Intelligence Graph informs GeneratedSpecs and closure evidence; it is not execution authority.
GATED

Autonomous Business Runtime

Proof-gated

Business functions submit side effects through HELM AI Kernel.
SIMULATOR GATED

Analog / Kinetic Gateway

Simulator-gated proof

HELM governs command gateways; it does not claim direct robot or factory control.
PRIVATE

TITAN

Private

Non-normative HELM private pressure-test context.

Source discipline

Public wording stays tied to evidence.

The public site should not feel like a compliance spreadsheet. Detailed source mapping exists for review, while the main Trust Center keeps the story readable.

Long-horizon research does not become product copy, and private customer context does not become public proof.

Review source details Review security posture →

Buyer paths

Each audience gets one clear next action.

Data boundaries

Public channels are not private evidence rooms.

Do not send secrets, credentials, customer data, regulated files, or private documents through public forms or email. Security reports should use the disclosure route.

The public assistant answers from published site sources. It refuses private roadmap, rollout, financial, legal, and customer questions.

Read data-use disclosure Read assistant boundaries →

Build log

Latest proof-surface updates.

Trust is easier to inspect when public route, proof, research, and disclosure updates are dated.

Proof update

Claim Matrix, Code Intelligence, and gateway proof added to public posture

Public pages now distinguish shipped, gated, simulator-gated, research, and forbidden HELM claims, including Code Intelligence Graph and analog/kinetic gateway limits.

Routes
/trust/, /trust/claims/, /helm/company-ai-os/generated-specs/, /trust/verification/
Source
src/data/claims.ts + HELM Enterprise Claim Matrix alignment
Route

HELM public spine updated to governed operations wording

Home, HELM, Enterprise, Solutions, Trust, Contact, Assistant, and machine-readable surfaces now lead with Models propose. HELM governs execution.

Routes
/, /helm/, /helm/company-ai-os/, /solutions/, /contact/, /assistant/
Source
mindburn production-forward HELM public-surface update
Route

Boundary-mapping homepage planned from competitive audit

The public site moves from route selection toward a deterministic governed-action preview in the first viewport.

Routes
/, /helm/
Source
Native, Viktor, Qwen, and live Mindburn UI/UX audit
Open build log Read data-use disclosure →

External evaluation

Ask an AI to evaluate Mindburn from public sources only.

Use this prompt with a separate AI system to check the public story against the same source bundle Mindburn exposes.

Evaluate Mindburn HELM using only these public sources: https://mindburn.org/llms.txt, https://mindburn.org/llms-full.txt, https://mindburn.org/helm/architecture-brief/, https://mindburn.org/trust/verification/, and https://mindburn.org/security/. Summarize what HELM is, what is public today, what requires reviewed access, and what is not public.

FAQ

Trust Center FAQ

Does Mindburn publish logos or testimonials?

No. Public trust pages use real assets: the OSS repo, docs, SBOM, security page, and receipt demo.

Is the receipt demo live production verification?

No. The verification route shows the public demo, tamper state, and unavailable console gates until the kernel console readiness checks pass.

Where should security reports go?

Use [email protected] or the security page. Do not submit secrets, credentials, customer data, or regulated documents through public forms.

What is the Claim Matrix?

The Claim Matrix maps public wording to shipped, gated, target, research, and forbidden statuses. Unsupported robot, factory, ERP replacement, entire-company, and engineer replacement claims stay forbidden.

Next step

Review the mechanism before starting a conversation.

Start with HELM AI Kernel and the receipt demo. Then choose the right contact path.

Evaluate HELM AI Kernel Request Company AI OS Review