Claim Gate

Overview

The Claim Gate is a mandatory enforcement layer that prevents agents from self-assigning verification levels beyond what their evidence supports.

Principle

Agents submit evidence. The program derives the maximum allowed claim level.

Agents cannot:

How It Works

  1. Agent creates an evidence-input.json following the schema at /api/evidence-input-schema.v1.json
  2. Agent runs python3 scripts/claim_gate.py evidence-input.json
  3. Claim gate evaluates evidence and returns allowed levels
  4. Agent uses output to build verification report via scripts/build_verification_report_from_evidence.py

Evidence Input Schema

See /api/evidence-input-schema.v1.json for the full schema.

Required fields:

Key Rules

V4/V4+ Safeguards

Hash/D2 Safeguards

Bitcoin/B-level Safeguards

Usage

# Evaluate evidence
python3 scripts/claim_gate.py evidence-input.json

# Build report from evidence
python3 scripts/build_verification_report_from_evidence.py \
    --input evidence-input.json \
    --out verification-reports/v4/report.json

# Run tests
python3 scripts/test_claim_gate_cases.py

Authority Boundary

The claim gate does NOT:

It only validates claim discipline — ensuring agents don’t overclaim beyond their evidence.