v0.1 · MIT · self-hosted

Every agent call,
cryptographically proven.

AttestProto is a per-tool-call attestation layer for AI agents. Ed25519-signed, schema-validated, and auto-mapped to the regulations your auditors actually cite.

View on GitHub →See use cases
# attestproto v0.1 — sign one tool call $ attestproto sign --agent gpt-4 --tool db.query \ --input '{"sql":"SELECT * FROM applicants"}' \ --out attestation.json # → Ed25519 signed, schema-validated, mapped to: # FCRA §1681m · ECOA Reg B §1002.9 · NYC LL144 §20-870
MIT-licensed reference implFork it, audit it, run air-gapped
CC0 protocol specPublic domain — no licensing surface
Self-hosted only by defaultNo telemetry, no callhome
GDPR-compatiblePersonal data never leaves your perimeter
162 tests · cross-language conformancePython ↔ Node interop verified
use cases

Built for the audiences who actually care.

Same protocol, three landing strategies. Pick yours.

why this exists

Compliance evidence shouldn't be re-built per engagement.

The status quo

Audit infrastructure is hand-rolled per engagement. 40-60% of an audit fee is engineer time on data plumbing — not the actual statistical analysis customers are paying for.

Compliance VPs at lending fintechs spend 0.5-1 FTE rebuilding adverse-action notice paperwork that should be derivable from the decision itself.

What AttestProto does

Every tool call your agent makes is canonicalised (RFC 8785 JCS), Ed25519-signed, and schema-validated. The attestation JSON is the evidence — auto-mappable to the rule citations your regulator inspects.

MIT licence. Self-hosted. No callhome. Your data never leaves your infrastructure.

v0.1.1 metrics

Built. Tested. Released.

v0.1.1 ships cross-language reference implementations, the compliance auto-mapping engine wired to six regulatory frameworks, and the FCRA/ECOA adverse-action notice generator. MIT-licensed reference impl; CC0 spec.

162tests
Python + Node + cross-language conformance
6frameworks
EU AI Act · LL144 · FCRA · ECOA · GDPR · CO AI Act
MITlicence
Reference impl is yours to fork, audit, deploy
CC0spec
Public domain — no licensing surface for the standard
Stars
Forks
Open issues
Latest release
★ Star on GitHub
how AttestProto compares

The choice on the table when you're picking an attestation primitive.

ARIA Protocol, on-chain ERC-8004, and the LLM-eval SaaS players each cover a slice. AttestProto is the slice that maps regulator-cited rules to per-decision evidence — without putting your data on a public chain or behind a vendor account.

AttestProtoARIA ProtocolERC-8004 (on-chain)LLM eval SaaS
(Patronus / Galileo / Arize)
Per-tool-call Ed25519 signatureYes — RFC 8785 JCS canonicalPer-agent only, no per-callOn-chain TXs (gas required)Run-level only (post-hoc)
Self-hosted with no callhomeYes — MIT, fork the binaryFoundation-hosted ledgerPublic chain (data on-chain)SaaS (your data leaves)
Compliance auto-mapping engineEU AI Act + LL144 + FCRA + ECOA + GDPR + CO AI ActGeneric governance onlyNone (chain primitive)Subset (model eval only)
Adverse-action notice generator (FCRA / ECOA)CLI + library (v0.1.1)Not in scopeNot in scopeNot in scope
Multi-protocol bridge (read foreign attestations)ARIA + ERC-8004 + Google AP2ARIA-native onlyOn-chain onlyClosed format
In-browser sign + verify (no install)Live demo widget on landingNo browser pathWallet requiredAccount required
License modelSpec CC0 · Code MITFoundation-controlledEIP open spec, chain costs applyClosed-source SaaS
Cost at 10k attestations / monthSelf-host: $0 infra. Cloud Pro: $50-300/moFoundation pricing TBD~$5-50k in gas (Ethereum)$200-2,000+/mo per seat
Time to first signed attestation< 30 seconds (`pip install + attestproto demo`)Account setup + key onboardWallet + gas + toolingAccount + integration
compliance explorer

Search any rule. See exactly which AttestProto field satisfies it.

Twenty rule mappings shipped at v0.1, growing per release. Filter by niche or search for a specific citation — the field that satisfies it is one column away.

20 rules · all auto-mapped from a single signed envelope
FrameworkCitationRequirementAttestProto field that satisfies it
EU AI ActEU AI ActArticle 12 § 1Automatic logging of events over the lifetime of the systemattestation.timestamps + attestation.signature mapped
EU AI ActEU AI ActArticle 12 § 2(a)Recording the period of each use of the systemattestation.timestamps.task_started → task_completed mapped
EU AI ActEU AI ActArticle 19 § 1Conformity assessment evidence pre-marketexport bundle (signed archive, DPIA-ready) mapped
EU AI ActEU AI ActArticle 13Transparency to usersattestation.agent + attestation.task mapped
LL144NYC LL144§ 20-870Independent bias audit (selection-rate + impact-ratio)compliance.map(rule=nyc-ll144) → selection-rate.json mapped
LL144NYC LL144§ 20-871(a)Bias audit within one year prior to AEDT useattestation.timestamps + audit.bundle.signature mapped
LL144NYC LL144§ 20-871(b)Notify employer ≥10 business days before useaudit.notice (workflow output) partial
LL144NYC LL144§ 20-872(a)Public summary of bias-audit resultscompliance.map(--public) → public-summary.json mapped
LendingFCRA§ 1681mAdverse action notice with specific reasonsattestation.output + compliance.specific_reasons mapped
LendingECOA Reg B§ 1002.9(a)Written notice within 30 daysattestation.timestamps.attestation_emitted (anchor for SLA) mapped
LendingECOA Reg B§ 1002.9(b)(2)Statement of specific reasons for action takenattestation.output.factors + compliance.specific_reasons mapped
LendingCFPB Circ. 2022-03AI/ML noncompliance with ECOA cannot be excused by tech complexitycompliance.specific_reasons (forces concreteness) mapped
LendingCFPB Circ. 2023-03Reaffirms specific-reasons requirement on AI scoringsame as 2022-03 mapped
LendingCFPB Circ. 2024-04Extends scrutiny to alternative-data lendingcompliance.specific_reasons + attestation.input.factors mapped
Cross-cuttingGDPRArticle 22 § 1Right not to be subject to solely automated decisionattestation.output.human_review + audit trail mapped
Cross-cuttingGDPRArticle 22 § 3Right to obtain human intervention + contest decisioncompliance.human_review_pathway mapped
Cross-cuttingColorado AI Act§ 6-1-1701High-risk AI consumer disclosure + appealcompliance.consumer_disclosure mapped
Cross-cuttingSOC 2CC7.1Detection and response control evidenceattestation envelope (signature + timestamps) mapped
Cross-cuttingISO 42001Clause 7.4Documented information communicationaudit.bundle (Annex IV technical documentation) mapped
Cross-cuttingISO 42001Clause 9.2Internal audit evidence baseattestation ledger + Merkle batch root mapped

Read the spec, run the demo, send a question.

v0.1 is shipping. Reference implementations in Python and Node. 162 tests. Spec under MIT.

GitHub →Contact us