Evidence, references and source custody
What does MeshQu do with the evidence behind a decision, and does it establish that the evidence is true?
MeshQu binds evidence by reference. It stores digests, references and metadata rather than the artefact bytes, and binding a digest does not establish that the referenced artefact is authentic, complete or the right one for the policy. A custodian outside MeshQu can sign for an evidence item. Where MeshQu does hold bytes, the custody routes prove byte identity against what was sent, and nothing about the source.
In more detail
The useful property here is narrow and worth stating exactly. A digest recorded at decision time fixes which artefact was considered, so a document produced later can be checked against what the decision actually used. That is a real answer to a real audit question. It is not an answer to whether the document was genuine. MeshQu hashes what it is given, and it cannot attest that the digests a caller submits reflect reality. Independent custody is where a stronger claim can come from. Custodian signatures are produced outside MeshQu, by banks, vendors or auditors, and MeshQu never holds their private keys, so a verifier supplies the trust roots out of band. Whether a deployment uses independent custodians at all is an operating choice rather than a product property. Where bytes are stored, the documentation is precise about what storage proves: that what was stored is byte-identical to what was sent, and that a later availability check reports what was retrievable at that moment.
Limits
MeshQu does not store evidence content. It stores digests, references and metadata.
Binding a digest does not establish that the referenced artefact is authentic, complete, or the right artefact for the policy.
The source-artifact record binds only the hash. Type, filename and size are stored but are not bound.
The source-custody and sealed-review routes have no availability assessment: the 17 July 2026 verified-claims register does not cover them. Availability on this entry is recorded as not publicly confirmed for that reason, and should be confirmed before either route is relied on.
Whether independent custody is used at all is a deployment and governance responsibility, not something the product supplies.
Where it applies
Answer an auditor who asks whether the signed record establishes that the underlying documents were genuine.
Show which artefact a decision considered, by comparing a document produced later against the digest recorded at the time.
Decide where independent custody of evidence has to sit before a stronger evidential claim is made.
Sources
MeshQu docs - Receipt reference (opens in a new tab)
The reference-based evidence model, the custodian-signature boundary, and the documentation owner's own statement of the copy traps around stored evidence.
MeshQu docs - Trust model (opens in a new tab)
The statement that MeshQu hashes what it is given and cannot attest that submitted digests reflect reality, and the list of deployment and governance responsibilities.
MeshQu docs - Source custody: finalise a staged upload (opens in a new tab)
What storing bytes proves, stated by the route itself: byte identity against what was sent, and nothing about the source.
MeshQu docs - Retrieve a sealed review submission and its canonical components (opens in a new tab)
How an availability state is produced, and why an unchecked entry is not a finding about the bytes.
Related answers
Decision Receipts: what the record contains
What is in a Decision Receipt, and what does it prove?
The limits of verification
What does verifying a Decision Receipt actually prove?
What a receipt evidences about a responsible-AI commitment
Can a Decision Receipt show that an AI decision was fair, accurate or correct?
Independent verification: the bundle, and what replay checks
How does someone independently verify a Decision Receipt, and what does replay check?