TaskQ audit reports package signed HashIQ entries, placement proofs, credential access events, registry snapshots, blocked attempts, settlement references, and external anchors into evidence a buyer can review, export, and verify.
The PDF is the human layer. The export bundle is the proof layer: signed chain entries, placement proofs, credential access events, registry snapshot proofs, attestation references, settlement journals, anchor proofs, and verifier instructions.
{
"report_id": "rpt_2026_06_12_ca",
"scope": { "tenant": "cus_...", "range": "184000-184999" },
"chain_entries": "entries.ndjson",
"placement_proofs": "placement.ndjson",
"credential_access": "credential-access.ndjson",
"registry_snapshots": "registry-proofs.json",
"settlement_journals": "settlement.ndjson",
"anchors": "anchors.json",
"verifier": "taskq-verify"
}Each row in the report maps back to a signed event. The buyer can inspect the summary, then drill into the exact proof behind any request.
| Evidence class | What it proves | Export material |
|---|---|---|
| Policy enforcement | Resolved policy, constraints, blocked attempts, no loosening | policy_hash + rejection entries |
| Placement | Chosen worker satisfied constraints; no cheaper compliant option skipped | placement_proof + registry snapshot |
| Credential access | Which fingerprint was used, where, under which custody class | credential.access events |
| Execution integrity | Canonical request/response hashes and terminal status | HashIQ chain entries |
| Settlement | Every billable request maps to a journal line | journal entries + reconciler output |
| Anchoring | The chain range was witnessed and cannot silently move | range root + witness signatures |
Verification replays the chain range, validates signatures, checks anchors, rebuilds placement decisions from registry snapshots, confirms credential access events, and reconciles settlement references.
When a request cannot meet jurisdiction, custody, attestation, capacity, or price policy, TaskQ returns refusal with evidence instead of silently relaxing constraints.
The pool was empty. The report shows the policy hash, empty eligible-set proof, and the constraints that eliminated workers.
The request tried to weaken defaults. The report proves customer defaults stayed stronger than a request-level override.
Compliant capacity existed but exceeded the price ceiling. The report shows floor-price evidence and why execution was refused.
Each closed billable request has exactly one settlement journal, and every journal points back to a valid chain entry.
TaskQ’s honest moat is exportable proof: the customer can leave with their evidence, but no competitor can backfill yesterday’s witnessed chain.