Zanii Proof

Seven jobs, one evidence foundation

Source modules are building blocks, not a count of hosted products.

CaptureControlOversightPrivacyComplianceCommerceProvenance

80 source modules; the private verifier build is excluded. Availability and test evidence are separate. No module is labeled hosted/live without recorded operational evidence.

Module and jobDeliveryVerification evidence
@zanii/a2a-directory

Agent discovery + trust lookup for A2A / agent cards - resolve an agent DID to its verified history, checked offline (not just the server's counts).

Usable source / SDKnot assessed
Hosted: unverified
@zanii/admissibility

Turn a Zanii audit bundle into a court-ready bilingual (Arabic/English) evidence pack for UAE Electronic Transactions Law 46/2021 - verification steps for a judge/expert + offline reproduce commands.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/age-assurance

Provable age-gate attestations without storing the ID - a real 2026 regulatory wave (UK Online Safety Act, EU, US state laws). Prove an age threshold was checked (over-18, over-13) by a named method/provider, committing only a salted subject reference and the boolean result - never the DOB, never the document. The provider check is INJECTED (age-verification service / document check / reusable ID): we prove the gate happened, you supply the trust root; it does not assert the person truly meets the threshold.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/agent-bank

Treasury accounts for AI agents: spending policy (per transaction, per day, per merchant category, allowed counterparties) checked before money moves, a double-entry journal that always balances, and reconciliation of the journal against settled payments. Zanii never holds funds.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/ai

Instrument AI-agent framework tools with Zanii in one line — every tool call becomes a signed, verifiable receipt. Works with the Vercel AI SDK and any tool with an execute() function.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/attest

Bind which code was running to a receipt: an Attestation claim + attestation_hash provenance field, with quote verification delegated to an injected platform verifier (SGX/Nitro/TPM). Honest by default — no verifier means quote NOT verified.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/broker

The credential broker that closes Zanii's stated custody hole: the agent never holds raw API keys - the broker does, and it injects them ONLY into calls that produced a receipt first. Receipt-or-nothing, fail-closed (ledger unreachable = no call), credentials bound to URL origins so they cannot be exfiltrated to attacker-chosen hosts, secrets never in hash preimages or logged responses.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/carbon

Compute-emissions receipts for AI: energy per model call from measured or declared power, grid carbon intensity by region and hour, and period totals ready for Scope 3 / CSRD reporting, with every figure traceable to its inputs.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/chain-bridge

One keypair, two products: bridge the Zanii L1 (blockchain.zanii.agency) to the Zanii transparency ledger. A chain wallet's Ed25519 pubkey IS a ledger did:key, so the bridge PROVES a registered chain agent's claimed DID controls its wallet (address = bech32 zan + blake3(pubkey)[:20]) and resolves it to its verified ledger history - the agent paying you on-chain is provably the agent whose actions you can audit.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/cli

The Zanii command line: generate identities, sign delegations, record + verify proofs, and export audit bundles, no code required.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/compliance

Turn a Zanii audit bundle into an auditor-ready compliance report - offline verification, action breakdown, anchoring coverage, and control mapping, rendered to Markdown.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/conformance

Conformance vectors for the Zanii protocol: canonical encoding, payload commitments, keys, delegation, receipts, Merkle proofs, signed tree heads and audit bundles, each with must-accept AND must-reject cases. Language-neutral JSON plus a runner.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/connectors

One-call proof recording for OpenAI function calls and Anthropic tool_use, plus a toolbox wrapper. Platform-shaped glue over the SDK's wrapTool. Also Gemini, Bedrock, the Responses API, computer use and code execution.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/constitution

Machine-checkable behavioral constraints as signed policy - publish an agent's rules (forbid / require_approval / allow over scope patterns), stamp the constitution hash on every receipt via the existing manifest_hash field, then PROVE each action was governed: a forbidden action with the stamp is a provable violation, and a rulebook change mid-stream is visible. Generalizes the fta/sca walls into a framework.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/core

Zanii protocol: verifiable identity, signed action receipts, Merkle transparency-log proofs. Pure, no I/O.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/credentials

Verifiable institutional credentials — diplomas, licences, certifications — offline-verifiable in milliseconds with no call to the registrar. Four checks: signature, expiry, revocation (with an explicit freshness window), and a domain-bound issuer root-of-trust.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/custody

Supply-chain provenance: co-signed custody chains for physical goods. Each handoff (factory→shipper→customs→retailer) is a 2-party co-signed receipt, multi-party events are N-party swarm receipts, and item history is a pseudonymous verifiable slice — no party trusts another's database.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/custody-verdicts

The one custody verifier: computes the eight subject-page verdict rows (purpose, self-sovereign, nobody-read-it, never-trained-on, no-copy-kept, residency, subprocessor, minimization) from an offline-verified SubjectHistory. Pure, synchronous, no network. The subject page and @zanii/cli call this same function.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/cv

AgentCV — a signed, portable CV over data the ledger already holds. The summary is the ledger's own un-inflatable aggregate; curated entries each point at a verifiable receipt/cert. Proves the entries are real, never that the history is complete.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/data-custody

Prove what happened to the DATA an agent touched - every touch of a user's document, where it went, and when the original was destroyed. NOT supply-chain @zanii/custody (physical goods). A custody record is a Proof receipt with the data ref as its subject_tag: same chain, same anchor, verified with verifyReceiptChain unchanged, read via the existing /v1/subjects/{tag}. Purpose is bound to the delegation (data.<class>.<op> targets), so a read-scoped agent physically cannot record an export. Hashes only; the party being audited is us.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/decisions

Gig/creator algorithmic fairness: decision receipts a worker can audit — completeness (every decision about me, tamper-evident), rule-consistency (the same rulebook decided everyone, via manifest_hash), and committed factors (no post-hoc rationalization). Never claims 'the algorithm is fair'.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/education

Integrity for AI in learning and assessment: AI-assistance declarations per assignment, provable submission times, assessment-rubric versions, tutor-session summaries and grade-change trails, with students identified by pseudonymous tags.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/elections

Chain of custody for election operations, not vote casting: sealed-container and equipment custody with co-signed handoffs, tabulation batches whose totals add up, and a registry of AI-generated political content with disclosure labels. Never handles ballot choices or voter identities.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/embed

Drop a "Verified by Zanii" badge onto any website - framework-free, zero-dependency SVG + snippet builders linking to public verify and agent-profile pages.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/energy

Agent actions on the grid, made provable: meter-signed readings, flexibility and demand-response dispatch with verified delivery, EV charging sessions, and renewable-origin claims that cannot be counted twice. Meter accuracy stays with the meter (trust root).

Usable source / SDKnot assessed
Hosted: unverified
@zanii/erc8004

Register & resolve agents as ERC-8004 on-chain identities wired to their Zanii proof endpoints - build the registration file, resolve one back, and verify the agent's history offline.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/escrow

Conditional payments between agents as verifiable agreements - co-signed terms, a deterministic release verdict when a matching receipt lands in the log with an inclusion proof, and signed release/refund instructions. Zanii never holds funds: money moves on x402/on-chain rails and settlement is verified, not custodied.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/federation

Cross-ledger interop that turns a transparency log into a protocol - published witness policies (an STH counts only when M-of-N independent witnesses co-sign it), split-view gossip across vantage points, and the equivocation proof: two signed STHs, same log, same size, different roots. Offline-verifiable, deterministically damning.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/fta

FTA filing evidence for Meezan & Books - receipt 'prepared a filing' with the wall enforced (prepares, never files/advises), and the Tax-Agent handoff as its own signed receipt. Zero deps.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/gateway

Drop-in proxy that receipts every HTTP API call an agent makes - zero code changes. Edge-compatible (fetch handler) plus a Node server and CLI.

Gateway requirednot assessed
Hosted: unverified
@zanii/gov

Public-sector algorithmic accountability: benefits, visas, fines, licences. A state decision with no governing rulebook cannot be constructed, and appealPack puts a court-ready, bilingual proof of what the algorithm did to you into the citizen's own hands.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/guard

Stop attacks on AI agents before they land, and prove what was stopped: an inline guard before every action (canaries, tool pinning, exfiltration sequences, taint from untrusted content, rate limits), an evolution tracker that holds high-risk actions until a changed agent is re-tested, and a sandboxed red-team harness. Decisions are signed receipts.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/health

Medical audit trails without leaking the pattern: per-episode unlinkable tags (the patient supplies the tag, never the seed), plus who accessed the record, which model recommended what under which protocol, and which named clinician took responsibility. Break-glass is loud, not impossible.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/identity-bridge

Consume enterprise agent identity as an input, don't compete with it. Bind a Microsoft Entra Agent ID or a SPIFFE SVID to a Zanii did:key with a two-way attestation - the external identity's subject and the did:key both sign the same binding, so a Zanii receipt is provably by the agent the enterprise already governs. Token validation is injected (the @zanii/attest discipline): we verify the binding, you plug in the Entra/SPIFFE trust root.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/insurance

Machine-verifiable underwriting for agent liability - risk FACTS from the ledger's own un-inflatable aggregate (never a score), a co-signed policy, and claims that resolve against Merkle-proven receipts instead of disputes. Zanii states facts; insurers price. Ledger silence is unknown, not good.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/kms

Stop storing agent private keys in plaintext. Seal a key or identity under a passphrase with scrypt + AES-256-GCM, unseal only to sign. Zero dependencies.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/kya

Know Your Agent: screen a counterparty agent against a deny-list / injected sanctions provider before your agent transacts, and receipt the check (kya.screening). Rides @zanii/a2a-directory; the list is injected.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/kyb

Owner-to-legal-entity binding - the missing link between an owner did:key and who is legally on the hook. Two-way: the company serves .well-known/zanii-owner.json on its own domain AND the owner signs an attestation naming that domain and entity; verification requires both. Zanii verifies domain control, never corporate registration - the claim becomes verifiable and attributable, not officially certified.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/langchain

Receipt every tool call in a LangChain or LangGraph run with one callback handler - no changes to your tools. Signed, verifiable Zanii proofs across the whole chain.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/mcp-proxy

Zanii MCP proxy: front any MCP server so every agent tool call is recorded to a Zanii transparency log, with zero agent code changes.

Gateway requirednot assessed
Hosted: unverified
@zanii/memory

Provable agent memory: a memory.write receipt that hash-chains what an agent remembered and when (salted content commitments, tamper-evident links) so an auditor can prove 'why did it decide that?'.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/minimization

Prove data minimization (GDPR Art. 5(1)(c)) - unprovable today, only assertable. Commit the declared MINIMUM field set, then prove every field an agent accessed was inside it; any access outside the declared minimum is a flagged violation. Composes @zanii/redact field commitments and purpose-bound delegation into a proof a regulator accepts. Proves declared-vs-accessed, not that the minimum itself was well-chosen.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/monitor

Independent Zanii log monitor: continuously verify the transparency log is append-only (consistency proofs) and that anchored checkpoints hold. Makes 'trust no one' operational.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/openai-agents

Receipt every tool call an OpenAI Agents SDK run makes, via lifecycle RunHooks - no changes to your agents. Signed, verifiable Zanii proofs.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/otel

Surface every agent action as an OpenTelemetry span - Zanii proofs show up in the observability stack you already run (Datadog, Honeycomb, Grafana).

Usable source / SDKnot assessed
Hosted: unverified
@zanii/payments

Correct, verifiable money for Zanii payment receipts - integer minor units (bigint), per-currency precision, never a float. Builds the validated payload for recordPayment.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/policy

Decide whether an agent action is allowed before it runs - amount caps, allow/deny lists, rate limits, and payload conditions that delegation scopes can't express.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/pq

Post-quantum migration rails - hybrid dual-signing (Ed25519 + ML-DSA-65, both must verify), a PQ key binding signed by BOTH keys (proves possession of each), and a transition record anchored into the Merkle log so the binding provably predates any future break of Ed25519. Old receipts are not re-signed: their post-quantum protection is the anchored timestamp, and we say so.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/provenance

AI-content provenance: sign an artifact's hash with the creating agent's did:key, anchor it on the ledger, and resolve the creator to their verified history — 'made by a known, accountable agent' vs 'appeared from nowhere'.

Usable source / SDKunit tested

Tested artifact 0.2.0
2026-10-24T11:39:50.621Z
Windows-local-source-tests
sha256:5823e64d3a0cd960…
historical snapshot


Hosted: unverified
@zanii/react

Drop-in React components that verify Zanii proofs client-side: VerifiedBadge, ProofViewer, AgentProfile, LedgerTicker. Show trust in your UI in minutes.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/redact

Selective disclosure for Zanii receipts - prove an agent acted on specific data without revealing the rest, via salted field commitments and Merkle inclusion proofs.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/regulator

A supervisor's toolkit for AI oversight: scoped live feeds of receipts, provably random audit samples (seed committed before the draw), incident-reporting deadline clocks for the EU AI Act, NIS2, DORA and GDPR, and signed supervisory findings and attestations. A labelled regime table (law, guidance or proposal, with sources) adds UAE PDPL, DIFC and CBUAE rows, and feeds can be scoped by jurisdiction binding.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/residency

Prove where an agent's data physically stayed - bind a region attestation to each data touch, confirm the attested region is inside a declared jurisdiction, and emit a receipt the moment it crosses a border. Answers 'prove it never left the UAE / the EU.' The region signal is INJECTED (cloud-provider region / TEE geolocation) - we verify the binding, you supply the geo trust root; the proof is only as strong as that source. A cross-border transfer rule table for UAE PDPL, DIFC and ADGM (labelled, not legal advice) checks each touch's destination.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/retention

Provable data-deletion attestations for GDPR Art. 17 - signed, timestamped receipts that reference the erased subject by a salted commitment, never raw PII.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/robotics

Receipts for embodied AI: owner-bound device identities, safety envelopes checked before a command reaches an actuator, device-signed telemetry, emergency stops, and a tamper-evident black box for incidents. Proves what the device signed and what the controller commanded; sensor truth stays with the hardware (trust root).

Usable source / SDKnot assessed
Hosted: unverified
@zanii/runtime

Zanii agent runtime: deterministic rails — the model proposes, tested code disposes. Scoped authority, earned status words, and a human gate for irreversible actions.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/sdk

Zanii client SDK: instrument any agent to record signed, provable action receipts, and verify proofs with zero trust in the server.

Usable source / SDKconformance double

Tested artifact 0.6.0
2026-10-24T11:39:49.225Z
Windows-local-source-tests
sha256:51576349c073f8c0…
historical snapshot


Hosted: unverified
@zanii/sentinel

Runtime behavioral monitoring for agents ('antivirus for agents'): score drift between what an agent actually does and what it was made to do, from the receipt stream — and the alerts themselves are tamper-evident receipts.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/signerd

Keep an agent's private key out of the agent's process. A local daemon unseals the key, signs receipts, owns the timestamp and the chain head; agents record through a socket with DaemonAgent.

Usable source / SDKconformance double

Tested artifact 0.2.0
2026-10-24T11:39:52.596Z
Windows-local-source-tests
sha256:581836b220da0b5c…
historical snapshot


Hosted: unverified
@zanii/sla

Machine-verifiable service agreements between agents - co-signed terms (scope, window, volume, max-gap uptime proxy) with compliance computed deterministically from receipts. The per-agent hash chain closes the omission hole: a broken prev-link is an INCOMPLETE_RECORD breach, so a provider cannot hide the bad hour. Disputes become arithmetic, and breaches cite receipts.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/subject

Per-subject auditability: end users hold their own key and independently verify what agents did on their account — a pseudonymous, platform-scoped subject_tag receipt slice — without seeing anyone else's.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/subprocessor

Prove the onward-transfer chain - which downstream processors actually saw the data, under which agreement (zero-retention, DPA id), tied to the junction/settlement. Answers the exact GDPR Art. 28 question every AI DPA raises: 'you sent my data to OpenAI, who sent it where?' A transfer to a party not in the declared subprocessor list is a provable, flagged violation. Proves recorded transfers; an off-record leak needs an attested enclave.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/succession

Agent lifecycle and key rotation with signed continuity - the owner attests 'B succeeds A', so reputation and CV history transfer verifiably instead of dying with a key. A compromise succession REQUIRES the compromise timestamp, bracketing the trusted window. Continuity of authority, not identity of code.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/swarm

N-party (3+) co-signed receipts for agent teams: genuine M-of-N Ed25519 threshold signatures over a canonical body (not a 2-party chain). Recorded on the ledger via POST /v1/swarm (SPEC §14) and verifiable offline.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/testing

Zanii test utilities: an in-memory ledger (with real Merkle proofs) plus identity/cert/receipt fixtures — unit-test instrumented agents with no server or network.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/trading

Accountable agent trading desks: pre-trade checks (limits, restricted lists, fat-finger bands) recorded before an order leaves, best-execution evidence across venues, clock-sync statements, and order-to-fill trails a compliance officer can replay.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/train

Training-data provenance: hash a training set into a sorted-Merkle manifest, then PROVE a named subject's data was absent from a real training run - a non-membership proof anyone can check offline. 'It was never used to train a model' becomes a yes, not a promise. Proves what was recorded went in; cannot prove an off-manifest copy didn't (state the enclave boundary).

Usable source / SDKnot assessed
Hosted: unverified
@zanii/vc

W3C Verifiable Credentials 2.0 bridge - export Zanii credentials and AgentCVs as spec-conformant VCs with genuine eddsa-jcs-2022 DataIntegrityProofs (Zanii already speaks did:key + Ed25519 + JCS, which IS that cryptosuite), and verify any did:key eddsa-jcs-2022 VC from anyone. Bridges the format, not the trust model: revocation freshness and domain binding remain Zanii-side checks.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/walls

UAE vertical 'wall' presets as enforceable, auditable artifacts - scopes never granted + a policy config + a manifest_hash + a receipted eval. SCA trading ('education only, never buy this'), RERA, TDRA, Legal, Consumer, DIFC/ADGM.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/webhooks

Zanii webhook receiver: verify the X-Zanii-Signature (HMAC-SHA256) and dispatch typed events. The safe, boilerplate-free way to react to agent activity.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/witness

An independent co-signer for a Zanii transparency log. Verifies a checkpoint is append-only, then counter-signs it so others can trust that tree state without running their own monitor. Speaks C2SP tlog-checkpoint and cosignature/v1, serves add-checkpoint and checks an n-of-m witness quorum.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/x402

Bind a Zanii payment receipt to its on-chain settlement - verify the tx a receipt references actually settled to the stated recipient. Chain reads injected; zero-dep JSON-RPC fetcher included.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/zk

Small, honest privacy primitives for Zanii: Pedersen commitments on ristretto255 with provable aggregates (a total or count opened while every individual value stays hidden), hash-chain threshold proofs (value >= t without revealing the value), and salted set membership. Offline-verifiable; no SNARKs, and each primitive states exactly what it proves.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/jurisdiction

Signed bindings of an owner or agent did:key to the regulator, jurisdiction, licence and sector it falls under, so a regulator (DFSA, CBUAE, ADGM-FSRA, SAMA, VARA, SCA, EU market surveillance authorities) can answer 'which agents are mine?'. Self-declared or regulator-attested, closed code registry, checkable offline.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/incident-report

Signed incident evidence into regulator reports with proofs attached: the EU AI Act Art. 73 serious-incident report (Commission draft template) and the OECD common reporting framework (non-binding), in a signed incident.report envelope verifiable offline.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/post-market

EU AI Act Art. 72 post-market monitoring and Art. 12 / 19 / 26(6) log retention evidence: a signed post_market.report per period with receipt counts, retention proofs, incident report hashes and caller metrics, verifiable offline.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/ai-register

A regulator AI-system register built from Ledger and Certify records: one entry per system (agent DIDs, owner, jurisdiction binding, Certify certificate, KYB record, descriptive fields), exported as EU AI Act Annex VIII, UK ATRS and Eurocities Algorithmic Transparency Standard field maps with null-with-reason for anything the inputs do not support, plus a signed zanii.ai_register snapshot and CSV / JSON helpers.

Usable source / SDKnot assessed
Hosted: unverified
@zanii/status-feed

One signed, append-only, hash-chained status feed per jurisdiction: delegation revocations, Certify certificate suspensions, revocations and reinstatements, incident references and jurisdiction bindings, each with evidence hashes. Append, verify (chain, signatures, time, gaps), read since a sequence number, render as Atom or JSON Feed, and anchor the head as a Ledger receipt payload.

Usable source / SDKnot assessed
Hosted: unverified

Blackbox and Certify

These are separate products. Their hosted gateway availability has not been verified by this deployment. Source packages and integration examples do not establish that a hosted workflow is live.