DEV Community

henyi-tdca
henyi-tdca

Posted on

we Built a Sovereign-Credit Settlement Framework for Agents with 168 CNY and 3.35B Tokens

We Built a Sovereign-Credit Settlement Framework for Agents with 168 CNY and 3.35B Tokens

TDCA — A Sovereign-Credit-Anchored Institutional Protocol for Agent Collaboration. Now Open Source.

TDCA — A Sovereign-Credit-Anchored Institutional Protocol for Agent Collaboration.

Settlement runs exclusively on e-CNY (legal tender, central-bank liability); collaboration value is anchored to taxation as the minimum observable utility (fiscal fact); cognitive assets gain legal force through the national trusted copyright chain / balance chain (judicial fact).

No platformization, no rent-seeking, no tokenization. Currently in institutional demonstration mode (simulated) — every data point carries a provenance label.

Repository: https://github.com/henyi-tdca/tdca-protocol

1. The Agent Economy Doesn't Lack "How to Pay" — It Lacks "How to Split, Tax, and Account"

By 2026 the agent protocol stack has stratified: communication is owned by MCP/A2A (Linux Foundation); payments exploded into AP2/x402/ACP. But every one of them answers only "how an agent pays" — single-transaction authorization and settlement anchored to stablecoins, card rails, or platform goodwill — private credit.

None answers three questions:

  • Five agents earn ¥100 together — how do they split it fairly? (Shapley allocation exists only in papers; NeurIPS 2025 Shapley-Coop is a research workflow with no protocol, no engineering, no attestation)

  • How are boundary violations punished? (negative-space rules, fail-closed circuit breaking — no protocol-layer implementation)

  • How is the process audited? (attestable chain, legally admissible records — no protocol-layer implementation)

2. What TDCA Is: Not "Another Multi-Agent Framework" — an Executable Constitution

TDCA (Trusted Digital Collaboration Architecture) is the institutional protocol layer for multi-agent collaboration: a constitution of sixteen articles, a Negative-Space Function Language (NSFL), an NCA attestation chain, and a CCP collaboration contract protocol — forged with 168 CNY, 3.35B tokens, 11 days, and 613 green tests.

It answers not "how to make agents run," but "how the value of agent collaboration is measured, allocated, audited, taxed, and entitled" — the engineering of the institutional layer.

3. The Three Sovereign-Credit Anchors: What Separates TDCA from Every Existing Protocol

Anchor TDCA Mechanism Hardness Source
Settlement: e-CNY (sole option) Real-state MOU settlement runs only on e-CNY Legal tender + central-bank liability + settlement finality
Utility: taxation MOU = tax_in + tax_out; collaboration value measured by fiscal facts National fiscal audit — the only metric that is state-verified, legally enforced, and tied to real value creation
Rights: national trusted copyright / balance chain NCA attestation on-chain + cognitive-asset copyright registration Legally conferred, not technically conferred — balance-chain attestations hold evidentiary weight in litigation

"Legally conferred vs. technically conferred" is the dividing line between TDCA and the entire Web3 paradigm.

Dimension-by-dimension scan (x402 / AP2 / ACP / A2A / MCP / Shapley-Coop / token-agent projects): after the three anchors are combined, TDCA has no global equivalent.

4. Engineering Proof: Not a Whitepaper — Runnable Code

  • 613 tests green (six-repo pytest rerun: 613 passed / 0 failed, attestation NCA-VERIFY-613-001)

  • Cost evidence: 168 CNY / 3.35B tokens / 13.8K requests / 11 days (platform billing screenshot) — 2–3 person-months in traditional development; three orders of magnitude cheaper

  • Sandbox cases [simulated]: The Art of War multi-agent governance run (3 rounds incl. sandbox failures and exits) + hotel off-season dynamic scheduling (Shapley allocation, MOU VAT ¥10.19 fully anchored)

  • Dual-protocol engine: NIA-MACM cognition layer + IP rights layer, 18 tests green + CLI end-to-end

  • First "institution-as-code" closed loop: a live case exposed the "calibrate BATNA to φ" flaw (positive-sum constructible) → DEF-SUNZI-01 rules chartered (R-01~R-05, human-signed) → φ<BATNA interceptor shipped as executable code (read-only interception + unverified-declaration fail-closed) — the flaw is closed end-to-end from institutional text to runnable code; losing parties can no longer obtain allocations through any computational path

5. Compliance Red Lines: The Institution Itself Refuses Tokens

Written into the NSFL negative-space list: no token, no public sale, no dividend promise, no quasi-token rewards, real-state settlement only on the e-CNY fiat rail. Compliance is not an add-on — it is a byproduct of the three-anchor design.

6. How to Join (TDCA Five-Element Collaborative Open-Source Community)

TDCA is not a user group; it is a network of contractors — membership is established by signing agreements (admission NCAs):

  • L0 Observer: read, rerun sandboxes, file issues

  • L1 Contractor: sign an admission NCA (machine-readable template: GitHub ID + base-protocol acceptance; tools/enforce_entry.py self-checks in minutes)

  • L2 Node: complete an accepted contribution → node NCA → participate in Shapley allocation (simulated)

Community governance itself runs on TDCA: DCD lodgment, NCA attestation, NSFL circuit-breaking — all public. The community is the proof.

7. Roadmap

Phase Time Deliverable
S0 Prep Weeks 1–2 Repo + admission mechanism + code of conduct + funding
S1 Positioning Month 1 pack/ institutional textbook + launch post
S2 Proof of Run Months 2–3 dual/ sandbox engine + tdca-mcp-bridge + CTS-L1 suite
S3 Cold Start Months 3–6 First third-party CTS-L1 conformance statement (north star) + three-anchor access talks

Join

Disclosure: the project is currently in institutional demonstration mode (simulated). All cases, figures, and roles herein are provenance-labeled demonstration data — nothing here constitutes investment advice, a promise of returns, or real-funds settlement.

Welcome to the TDCA Five-Element Community. Here you are not just forking code — you are contracting into a new kind of digital-economy compact.

Top comments (0)