CAT-ALIGN-001: align AuthContract with controlling Veraxis category thesis

[veraxis-protocol]

Scope

Documentation only, additive. One section added to README.md, placed after Current status. Nothing removed or reworded.

• README.md — 16 insertions, 0 deletions

A placement decision worth reading

The work order asks for a standard role block. This repository has an owner-approved constraint that conflicts with putting it at the top, and that constraint wins.

docs/DEVELOPER-LANGUAGE.md records the AC-021 developer-language freeze: deep ontology vocabulary is introduced only after the developer-facing workflow is already clear. tests/test_veip.py::test_a11_readme_matches_da647ac_exactly and its CLI twin enforce this mechanically — the terms OIC, ZTL, OAM, VEIP and AEP must not appear before the ## Quick start heading.

A first attempt placed the block near the top and failed those two tests. The tests were not changed. The block moved instead, to sit alongside the status and boundary material where the deeper vocabulary already belongs. The section says so explicitly, so the placement reads as intentional rather than accidental.

This is the work order's own rule applied: no test expectation changes caused merely by messaging.

What the section says

It names the category, this component's role, what must exist upstream, what it serves downstream, and the canonical thesis — then draws the boundary that matters for a receipt-issuing system:

A receipt records that a specific rule artifact, at a specific version, authorized a specific action, and lets a third party recompute that binding without trusting the issuer. That is evidence about an authorization. It is not the origin of institutional authority: recomputing a receipt establishes the integrity of the binding, not that the underlying rule was validly admitted by an institution entitled to establish it.

What is explicitly preserved

• The controlling product sentence and the developer spine, untouched
• **Implementation status:** experimental reference implementation (TRL 4), and its position before the first worked example
• "the automated natural-language source-to-rule comparison … is target behavior and is not yet implemented end to end"
• "The current code tests the mechanical trust chain for one synthetic banking specimen"
• The full "does not currently claim" list — production readiness, universal policy correctness, general legal correctness, automated NL source-to-rule comparison, production-grade institutional identity or PKI, universal runtime integration, broad professional usability, market adoption
• "AuthContract is not a policy engine", the verifier-established evidence-class wording, and every claim-ceiling phrase the A11 test binds

Claim impact

None. TRL/experimental status is preserved verbatim and the added text is restrictive in effect. No adoption is asserted for any party.

No runtime change

No code, schema, fixture, CLI behavior or test was changed. The diff touches one Markdown file.

Validation

Isolated Python 3.12.3 venv, pip install -e ".[test]" -c constraints.txt, plus setuptools>=61,<85 as CI installs:

Gate	Result
make test	343 passed, 0 failed
make falsify	4/4 bounded outcomes observed
make no-network	PASS — install/import/tests/CLI completed without outbound Python sockets

Intermediate finding, disclosed: with the block placed at the top, make test reported 2 failed, 341 passed — both the A11 README-ordering assertions. After repositioning, 343 pass with no test edited.

• Base SHA: 6c677aec730bff79dfc60a88d1721564000ad111
• Head SHA: 1dfeef7267d66fb8dc16bc37de6aedd62906c133

Controlling thesis

veraxis-protocol/institutional-continuity#2 — Veraxis Category Thesis v1.0. Canonical location after merge: https://github.com/veraxis-protocol/institutional-continuity/blob/main/THESIS.md

---

Generated by Claude Code