A specification document package, distributed free of charge. The Truth Arbitration Agent decides whether a claim's origin and vetting status qualify it for use in a governed decision — not whether the claim is true — and records sufficient evidence for every determination to be reconstructed offline.
When AI systems cannot explain which claims they trusted, every consequential decision is indefensible.
This package is a self-contained, normative, execution-neutral single agent specification, delivered as-is, sufficient for an engineering team to build the agent without further design decisions about agent logic, evidence shape, validator behavior, or failure semantics.
This package is not software, SaaS, a managed service, consulting, support, managed operations, or implementation delivery. It does not approve source semantic content, activate tool-state without a separate gate, relax policy, or perform persistent writes during startup verification. It is not a live runtime — nothing in the package executes; everything in the package describes how a conforming executable agent must behave.
4 structural properties, defined across spec §2.2, §6–§9, and §12, and elaborated through the failure semantics, workflow, gate logic, and evidence contract sections:
Every claim classified before admission. No claim can influence a governed decision without first being classified by origin. Claims from different origins are subject to different admission rules. The classification is mandatory, recorded, and enforced — not advisory.
Every verdict auditable. When the agent admits, denies, or blocks a claim, it records enough evidence for the determination to be fully reconstructed offline — by an auditor, a regulator, or a forensic investigator — without access to the system's runtime state.
Contradictions caught before the decision. When claims from different origins contradict each other, the agent detects the conflict before the claims can influence the governed operation. The verdict reflects what was found — and what was not checked.
Failures blocked, not degraded. When admission requirements cannot be met, the system blocks progress and escalates. It does not silently degrade. It does not permit the decision to proceed with reduced confidence. It stops, records why, and specifies what is required to resume.
These are properties of the specification. They are not warranties about a delivered running system.
| Truth_Arbitration_Agent_Specification.md | Normative specification. The authoritative document. |
| Truth_Arbitration_Agent_Semantic_Index.md | Semantic Index — Human Edition (D1). Canonical glossary of all 167 normative terms. |
| Truth_Arbitration_Agent_Semantic_Index.yaml / Truth_Arbitration_Agent_Semantic_Index.json | Semantic Index — Structured Edition (D2). The same glossary as data, 1:1 with D1. For tooling. |
| Truth_Arbitration_Agent_Coverage_Matrix.md | Semantic Coverage Matrix (D3). Links term categories to the specification's sections. |
| Truth_Arbitration_Agent_Ambiguity_Register.md | Ambiguity Register (D4). 15 lexical-overload resolutions. Read before normalization questions arise. |
| Truth_Arbitration_Agent_Conformance_Tests.md | Conformance tests you run against your implementation, with fixtures, oracles, and minimal run note. |
| Truth_Arbitration_Agent_Whitepaper.html | Public-facing positioning. Non-normative. |
| NNT_Companion/ | Non-Normative Tests — demonstration-only evidence that the specification is derivable. Not a reference implementation, and not part of the specification. Do not read it before implementing: read it only after your implementation passes the Conformance Tests. |
| Truth_Arbitration_Agent_README.md | Orientation and reading order. |
| Truth_Arbitration_Agent_License.md | The license. |
| Truth_Arbitration_Agent_Manifest.sha256 | SHA-256 integrity manifest. |
| verify.md | Self-audit runbook you run against the package you hold. |
Per spec §16 Reuse and Adaptation Guidance and §15 Conformance Conditions, 5 implementation deliverables your engineering team builds from this specification:
The executable that realizes the §5 state model, the §6 decision table, the §7 validator catalogue, the §9 gate logic, and the §8 evidence contract.
Binding the bounded parameters the specification deliberately leaves to your declaration: deterministic canonicalization rules and semantic_normalization_profile (§2.1, §13); concrete reason_code naming and machine-readable form serialization (§16.2); the V-006A severity threshold (§6 DO-005); the V-006B confidence threshold and conflict-formalism scales (§6 DO-006, §13); the readmission_budget value and clearance criteria (§12.1); and authorized external-actor identities for ESCALATION_REQUEST and CLEARANCE_RESPONSE (§10, §12).
Explicit serialization or snapshot-at-intake plus compare-and-commit (per §11 and AMB-011); snapshot freeze at intake; V-007 drift detection and restart under the new snapshot.
Sufficient for offline reconstruction — every emitted DECISION, FAIL_CLOSED, ESCALATION_REQUEST, CLEARANCE_RESPONSE, CHECKPOINT, and ROLLBACK record retained per §5.2, §8 and §10, replayable without runtime state.
The Conformance Tests run against your implementation under your declared policy profile, attesting all runnable and parameterized tests and confirming the declarative integration check.
The specification is complete enough that none of these requires interpretation of ambiguous prose. Spec §15 Conformance Conditions sets the five acceptance criteria your implementation must meet; the package README numbers them IAC-1 through IAC-5 for reference.
This pack declares its own numbers — how many artefacts it ships, how many canonical terms it indexes, how many conformance records it carries, what those records return. This runbook has you reproduce those declared values on your own machine, before you build anything against the specification.
Each check has the same shape: find the value the pack declares, observe the value your machine produces, record whether they match. A mismatch is a fact you report, not a judgment you make: either the bytes you received are not the bytes that were published, or your environment differs. Both are worth knowing before you build. Neither requires you to decide what it means.
Integrity check, run inside the unzipped package folder.
sha256sum -c Truth_Arbitration_Agent_Manifest.sha256
shasum -a 256 -c Truth_Arbitration_Agent_Manifest.sha256
Get-Content Truth_Arbitration_Agent_Manifest.sha256 | ? { $_ -notmatch '^\s*#' -and $_.Trim() } | % { $h,$f = $_ -split '\s+',2; if ((Get-FileHash $f.Trim() -Algorithm SHA256).Hash -ieq $h) { "OK $f" } else { "FAIL $f" } }Recompute the Manifest's own hash and compare it to the manifest_ref bound in the License.
The Conformance Tests document contains 27 test records anchored 1:1 to the specification: 25 directly runnable on the agent's projection surface, 1 parameterized against operator-declared inputs, and 1 declarative check verified at integration level. Each record cites the specification authority that grounds it, the input fixture that drives it, and the oracle that defines pass and fail. The Minimal Run Note declares the projection surface, input types, assertion result types, and the conformance condition for the suite as a whole.
Run the suite against your implementation. A fully conformant implementation passes the 25 runnable tests, satisfies the 1 parameterized test under your declared parameter envelopes, and attests the 1 declarative check at your release-acceptance gate.
Your implementation's conformance is established by the Conformance Tests, not by verify.md.
Five paradigm-isolated reconstruction contexts of the same model independently built the Truth Arbitration Agent from the specification pack alone — in Python, Rust, Prolog, C, and Bash — and ran the agent's 27 Conformance Tests records against each reconstruction, under live toolchains with hash-attested binary provenance.
All 27 of 27 conformance records converged across all five paradigms, with zero divergent records. The convergence comprises 26 convergent PASS and 1 convergent SKIP (record TAA-PT-020-DECL). The Minimal Run Note lists TAA-PT-020-DECL among the bounded external obligations not locally runnable: ordering enforcement is distributed across §11 Temporal/Ordering Contract and §9 Gate Logic, and is satisfied at the runtime layer through serialization or snapshot-at-intake plus compare-and-commit (AMB-011). Per paradigm, each of the five reconstructions' own run reported the identical profile: 26 pass, 0 fail, 1 skip, 0 error.
The five contexts were isolated from one another; the reconstruction prompt-author, model and training data are shared across all five paradigm reconstructions, as the companion declares.
This is evidence of derivability and self-sufficiency of the specification, not a certification, not a quality score, and not a set of implementations to follow.
The baselines/ are study artifacts produced under restricted conditions; they are not reference implementations and must not be used as implementation patterns.
Every figure here is bounded to this single NNT run on agent pack TAA-TRUTH-ARBITRATION-001. The NNT companion is not part of the specification. Do not read it before implementing: read it only after your implementation passes the Conformance Tests.
This supplement records the Truth Arbitration Agent's finite projection and its public conformance suite across Common Lisp, Brainfuck, spreadsheet formulas, an Analytical Engine-style program, and AWK. Rust is included as a separate comparator role. The supplement Manifest is anchored to the Manifest of the canonical TAA package. The canonical TAA Specification, Conformance Tests, Semantic Index, Coverage Matrix, and Ambiguity Register remain authoritative.
The projection evaluates next_state(current: WorkflowState, trigger: TransitionTrigger, validator_passed: bool) -> WorkflowState over 12 WorkflowState variants, 29 TransitionTrigger variants, and both Boolean values, for 696 tuples. Each tuple produces a canonical workflow state or E. E is a reporting token for a tuple whose result is not determined by the integrated canonical corpus; it is not a workflow state, verdict, action token, or gate class. This is a bounded state projection, not a full-agent or deployment evaluation.
The GNU Awk run produced 26 PASS, 0 FAIL, 1 SKIP, and 0 ERROR. The workbook, after a full recalculation in native Excel, reports 26 PASS, 0 FAIL, 1 SKIP, and 0 ERROR, matching AWK for every record. TAA-PT-020-DECL is SKIP because Conformance Tests §1 and §4 define distributed ordering enforcement as a runtime obligation that is not a locally runnable assertion.
The historical framing spans the Analytical Engine design described in 1843 and Rust's public announcement in 2010: 167 years between designs represented by contemporary renderings. Lisp (1958), rendered here in its Common Lisp dialect (standardized by ANSI in 1994), and Brainfuck (1993) supply additional points in that comparison. It does not imply that the Analytical Engine itself was built or executed in the nineteenth century.
The rule-by-rule source traces and E treatment are in SOURCE_RULES.md.
All rights reserved. You may download, inspect and run these files to reproduce the reported results. They are provided as is, without warranty, and are run at your own risk. No other right is granted.
This package is distributed free of charge. No payment is required, now or ever, for anything this license permits.
Subject to the Restriction section, you are granted, free of charge and perpetually:
You shall not:
The line. Using an implementation as internal machinery inside your own product or service is permitted, including commercially and at any scale: the third party consumes your product, and neither selects nor obtains the agent as such. Offering the capability itself is not permitted. Permitted: your application uses the agent internally to validate state, and your customers buy your application. Not permitted: you expose the agent as an API, ship it as a library, or offer it as a service, so that a third party can obtain or invoke its capability as a thing in itself.
The implementations you build are yours. SEMANTICHOR claims no ownership over the source code, builds, or deployments you produce from this specification.
The license sets out what is free; the reserved uses are not licensed for this package.
Inquiries about reserved uses: contact@semantichor.com
Where you obtained the package otherwise than at the point of distribution, you accept this license by accepting it at semantichor.com, in the same two confirmations, before any use of the package.
This is a summary. Where this summary and the license appear to differ, the license controls. Read the full license →