UNI Universal Natural Intelligence

Wiki · Hierarchical Active Inference

Claim-Guard Report

Hierarchical Active Inference · hierarchical-aif/reports/CLAIM-GUARD-REPORT.md @ b909801f3db4 (hierarchical-aif/motor-stack) — opens the published snapshot 8b4b5935bcba

How to read this page

A Plain and a Clear version of this page have not been written yet. What follows is the document itself.

Precise — the source document

This is the document. Rendered from the repository at the commit above, with nothing rewritten for the web. A gate re-renders it on every deploy and fails the build if a single byte differs.

Gate: H-AIF-G9 · Run: 2026-07-21 · Scanner: src/motor_stack_aif/claim_guard.py

Mechanical clamp on forbidden claim language. A phrase counts as a violation unless it appears in a negated context or inside a forbidden-wording catalogue.

Scope

directory files scanned violations
hierarchical-aif/reports/ 11 0
hierarchical-aif/docs/ 1 0
hierarchical-aif/ledgers/ 2 0
hierarchical-aif/protocols/ 2 0
hierarchical-aif/src/ 8 0
hierarchical-aif/tests/ 7 0
hierarchical-aif/scripts/ 2 0
total 33 0

Excluded, with reason

file reason
src/motor_stack_aif/claim_guard.py Holds the phrase list as its FORBIDDEN data. Self-scanning a dictionary is a definitional false positive.
tests/.../test_claim_guard_forbidden_phrases.py Uses the phrases as fixtures to prove the guard fires.
reports/CLAIM-GUARD-REPORT.md This file. A report that quotes offending phrases would flag itself, and a FAIL listing would then re-flag on the next run.

These exclusions are recorded so they are auditable rather than silent. Two self-reference artifacts were observed while building this guard: an unfiltered scan reported 19 hits (all inside the guard's own dictionary and fixtures), and a first FAIL report that quoted its findings verbatim then re-flagged them on the next pass, reporting 124. Both are scanner self-reference, not claims.

Result: PASS

No forbidden claim language in any substantive hierarchical-aif artifact.

22 phrases are clamped.

Limits

  • Semantic equivalents are not caught. A document could assert parity in novel wording and pass.
  • The negation heuristic inspects 180 characters of left context; a distant negation may be missed, and an unrelated nearby negation could mask a real violation.
  • The guard checks wording only, never whether the evidence supports a claim. Passing is necessary, not sufficient; that judgement stays with the reviewer.

sha256 9b71026c5a8d5579 — of the original file, so what was ingested stays checkable.