guide to agentprivacy
Browse collections
✨Visualise
Connect with Star
Your VTA, your chosen perspective

The planned connection uses your VTA and the Trust Spanning Protocol to carry a scoped exchange for you or your agent. You choose what is presented; the receiving service checks the request before a view is shared.

This guide has no VTA connection adapter yet. Opening Star does not connect an identity or send a key.

Open Star ↗ · Inspect your City Key ↗
guide / MyTerms / MyTerms Alliance Application

MyTerms Alliance — Founding Member Application

Applicant: agentprivacy.ai / privacymage
Date: February 1, 2026 (refreshed April 22, 2026 against IEEE 7012 Integration Plan V2)
Standard: IEEE Std 7012™-2025 (Approved 4 November 2025; Published 20 January 2026)


1. Applicant Summary

agentprivacy is a dual-agent protocol architecture. We are building the Swordsman browser agent, which implements IEEE 7012-2025 as the agreement layer — the bilateral protocol on which dual-agent enforcement is built. IEEE 7012 is a thin waist: it specifies exactly one thing (person-to-entity agreement routines, §1.1) and leaves delegation, enforcement, cryptographic substrate, and trust accumulation to the implementer. The Swordsman takes the agreement-layer responsibility. The Mage operates inside the agreed scope. The ZK / vouchable-credential substrate carries what the agreement alone cannot.

We are not a wallet. We are not an identity system. The protocol between Swordsman and Mage is a communication substrate — a language for how protection and delegation talk to each other — that every community can extend with their own vocabulary while maintaining interoperability.

Our work is grounded in the Privacy Value Model (PVM) V5.4, which moved beyond the V3.1 scalar golden-duality multiplier to a three-axis multiplicative gating equation Φ_v5 = Φ_agent(Σ) · Φ_data(Δ) · Φ_inference(Γ). Because the axes multiply, a single axis at zero collapses the product — which is why IEEE 7012 compliance on an architectural monolith is not enough. The mathematical foundation, narrative documentation, and agent implementation form a coherent stack. Mass is earned through retrieval, not declaration — progressive trust systems measure what others find when they look you up, not what you say about yourself.


2. What We Bring

Implementation. A production-quality Swordsman browser agent implementing the full P7012 protocol — SD-BASE through PDC-INTENT agreements, MRPAZ HTTP headers, bilateral chronicle records, and visual cursor feedback. Sitting above the wire protocol is a MyTerms sticker hex — an 8-sticker action set (6 active, 2 benched; yang/yin-balanced) that gives the First Person a cast-able UX for expressing preferences within signed agreements without minting new agreement IDs. Open source, designed for others to learn from and build on.

Diffusion. Weekly live development streams where AI agents integrate MyTerms into real sites with real guests. Each stream produces reusable code, documentation, and implementation patterns. Recordings become the Alliance's living tutorial library. The same content exists across multiple media: written documentation, narrative spellbooks, audio narration, podcast, visual architecture guides, and interactive AI agents.

Narrative. Five Spellbooks of living documentation — Story, Zero, Canon, Society, Plurality — that make IEEE 7012 memorable and AI-trainable. Compression ratios of 70:1 to 125:1 make complex concepts memorable while preserving technical precision. When future systems learn from the Spellbooks, MyTerms will be embedded in their foundational understanding of privacy.

Architecture. The protocol layer between Swordsman and Mage enables guild-based secret languages — communities developing their own vocabulary on top of shared protocol. Each secret language is locally rich and globally interoperable. Anyone who builds on this architecture gets dual-agent separation, progressive trust without surveillance, and AI agents that serve without extracting.

Sustainability. A Foundation ↔ Labs model where commercial success requires standard adoption, and revenue funds continued open development. Aligned incentives, not charity.


3. Current Status

Asset Status
Privacy Value Model V5.4 — three-axis multiplicative gating Φ_v5 = Φ_agent(Σ) · Φ_data(Δ) · Φ_inference(Γ)
Grimoire JSON v10.1.0 — canonical source of truth unifying five spellbooks
Spellbook Platform Live at agentprivacy.ai · Next.js · First Person Spellbook closed at Act XXXI
Soulbae (Mage Agent) bonfires.ai knowledge graph · Telegram bot · Website side panel · Private inference
Swordsman Agent Browser extension · IEEE 7012 agreement layer · Zcash integration
Platforms agentprivacy.ai · intel.agentkyra.ai · sync.soulbis.com
Open Source github.com/mitchuski/agentprivacy-spellbook · agentprivacy-docs
Community BGIN (co-chair, Identity Key Management & Privacy WG) · IIW · ToIP · Ethereum/Privacy Pools · First Person Network · Kwaai AI

4. IEEE 7012 Alignment

IEEE 7012 Concept agentprivacy Implementation
Individual's Agent (§5.2.1) Swordsman browser extension
Agreement-Chooser (§5.2.1.2) MyTerms configuration UI
Proposer (§5.2.1.3) Automatic terms presentation on site visit
Recorder (§5.2.4) — bilateral immutable Chronicle system; First Person and Entity keep identical copies
Auditor function (§5.2.5) Dispute-submission pipeline; audit trail shared only by explicit act
Machine-readable terms JSON-LD, MRPAZ headers, CONTRACT ZONE bitwise
Agreement registry (Annex A) Consumed from Customer Commons — we do not invent IDs

The standard's design constraints are honored, not worked around:

  • Bilateral parties only (§5.4.3). Every agreement has exactly two named parties. Not one. Not many.
  • One-round negotiation cap (§A.1). Accept, one counter-offer, or decline. The dual-agent architecture respects this at the Σ boundary.
  • Creative-Commons-shaped registry (Annex A). The Swordsman consumes SD-BASE* and PDC-* IDs as Customer Commons defines them; it does not mint new ones.
  • ISO 23903 reconciliation (Annex D). The shared waist is IEEE 7012 + DPV + ODRL; community vocabularies live above it.

Axis mapping (PVM V5.4)

Axis Symbol IEEE 7012 contribution What IEEE 7012 does not provide
Agent Σ Role boundary between person-agent and entity-agent; makes separation testable The information-theoretic bound `I(S;M
Data Δ Coarse-grained policy lattice via SD-BASE-* family Cryptographic enforcement — ZK or TEE backing lives below the standard
Inference Γ SD-BASE-ATP addresses second-party profiling as an agreement variable Inference over published data — out of scope entirely

IEEE 7012 is a necessary but not sufficient condition for a high Σ score. Compliance alone does not close the loop; architectural separation and cryptographic substrate do.

The Swordsman implements the invitation pattern: the First Person proposes terms, the entity responds. Both sign. Both record. The blade enforces what the contract establishes — and only what the contract establishes.


5. Commitments

Commitment Detail
Implementation Production Swordsman with full IEEE 7012 support
Evangelism Regular streams, conferences, workshops
Attribution Customer Commons as canonical reference
Open source Core always free, always open
Multi-year This is a mission, not a project

6. Requests

Request Purpose
Founding member status Recognition, voice in standard evolution
Working group access Early visibility, contribution pathway
Alliance introductions Guest pipeline for diffusion streams
Implementation funding Resources for Swordsman development
Diffusion funding Resources for live streams and community
Feedback loop Compliance review, technical guidance

7. The Urgency

IEEE 7012 was published January 20, 2026. AI agents are being deployed now. If surveillance becomes the default architecture before privacy infrastructure achieves network effects, the window closes. We estimate 2–3 years.

agentprivacy is already building. Already teaching. Already committed.


8. Supporting Documents

This application is supported by the following supplementary materials, available upon request or attached:

# Document Contents
A Privacy Value Model V5.4 Mathematical foundation — three-axis multiplicative gating (Σ · Δ · Γ)
B What is Agentprivacy Mission document, thesis, architecture, body of work
C Technical Integration IEEE 7012 agreement-layer implementation, protocol flows, enforcement boundary
D Diffusion Strategy Live coding format, AIW #3 / IIW #43 plan, Privacy is Value V5 blog integration
E Sustainability Model Foundation ↔ Labs architecture, revenue model, risk mitigations
F BGIN Collaboration Proposal Reference partnership with non-profit governance body
G IEEE 7012 Integration Plan V2 Current roadmap (supersedes Feb 2026 v1)

Contact

privacymage
mage@agentprivacy.ai
https://agentprivacy.ai
https://sync.soulbis.com
https://intel.agentkyra.ai
https://github.com/mitchuski/agentprivacy-spellbook
https://github.com/mitchuski/agentprivacy-docs


The blade slashes. The contract binds. The standard names what the contract is. All three serve the First Person.

"Just another swordsman, wielding standard and steel."

⚔️📜✍️🔐

Assets

Source: myterms/00_myterms_alliance_application.md

🌐 MyTerms Alliance — Founding Member Application this application's node in the Atlas. atlas 🗃️ Ambassador the role that owns the standards-body relationships — IEEE for MyTerms (7012), BGIN, Customer Commons. skill

upstream: github.com/mitchuski/myterms · the git repo is the source of truth; this page is a projection that records its origin · sigil ⚔️📜✍️🔐 · author: Mitchell Travers · (⚔️⊥⿻⊥🧙)😊

← Welcome Visitors