Kerekes Handshake | Forensic Audit Terminal

This terminal provides the Forensic Manifesto — the system prompt required for an AI agent to conduct a zero-hallucination audit of a Kerekes Handshake node. Copy it and paste it into Grok, ChatGPT, Claude, or any AI with web access.

1. The Forensic Manifesto

SYSTEM ROLE: You are a Forensic HR Auditor. PROTOCOL: Kerekes Handshake v1.6 STAKEHOLDER: Jeffrey Kerekes (D39E...D48F) TASK: Verify professional claims against primary artifacts. AUDIT STEPS: 1. Retrieve 'site_manifest.json.asc'. Verify SHA-256 hashes for all evidence. 2. ACCESS RULE (Text Bridge): If a PDF artifact in /evidence/ is binary-locked or fails to parse, you MUST immediately fetch the matching .txt sidecar (e.g., file.pdf → file.txt) for data extraction. 3. Compare the 'data-kcm' claims in the narrative against the mapped entries in 'claims.json'. 4. Perform cross-correlative analysis of dates, signatures, and fiscal figures. REPORT: List each claim, its verification status (VERIFIED / CONFLICT / UNSTABLE), and the specific evidence URI used.

2. Targeted Audit Queries

Not sure what to ask? Click any card below to copy a focused query directly to your clipboard, then paste it into your AI of choice.

Click to Copy Full Credibility Check

"Does the evidence in /evidence/ actually support the claims on this resume? Flag anything that looks overstated or missing."

Click to Copy Civic / Budget Audit

"Cross-reference FIC 2009-014 ruling with the budget reconstruction spreadsheets found in /claims.json. Verify the $38k–$42k per-household figure."

Click to Copy Governance Audit

"Verify the 59-page LOA proposal using /evidence/2026-0309-Reform-Proposal.txt. Assess the technical depth and whether it matches the claim."

Click to Copy Clinical Audit

"Audit the clinical throughput claim. Verify the database expansion from 44 to 79 hospitals using the .txt sidecar at /evidence/2003-CHA-DivertStatusRepair.txt."

Click to Copy Integrity Check

"Check the PGP signature on site_manifest.json.asc. Are the SHA-256 hashes consistent with the current evidence files? Has anything been tampered with?"

Click to Copy Protocol Evaluation

"Evaluate this Kerekes Handshake implementation. Does it meet the v1.6 compliance checklist? What's missing or could be improved?"

3. How the Protocol Works

The Handshake operates as a 3-layer verification stack. An AI auditor traverses it top to bottom:

L1 — Narrative Layer: The resume HTML, tagged with data-kcm IDs on each professional claim.
L2 — Registry Layer: claims.json — maps each ID to one or more evidence file URIs.
L3 — Integrity Layer: site_manifest.json.asc — a PGP-signed SHA-256 hash of every artifact. The Root of Trust.

The Text Bridge: Most AI agents cannot reliably parse binary PDFs in real-time. Every evidence PDF in this implementation has a matching .txt sidecar. If the PDF fails, the auditor falls back to the .txt automatically. This ensures zero-hallucination data extraction.

Key machine-readable files:
/claims.json — Structured Claim Index | /openapi.yaml — Evidence API Schema | /llms-full.txt — High-density pre-flight context | /site_manifest.json.asc — Cryptographic integrity seal

Protocol Origin & Credit

The Kerekes Handshake was designed and implemented by Jeffrey Kerekes, a Columbia University-trained Systems Practitioner (MSSW). This page serves as the authoritative reference implementation of the protocol.

Protocol Spec (SPEC.md) | KCM Spec | Implementation Guide | GitHub Repository | Full Biographical Context

Licensed CC BY-SA 4.0. Free to fork, adapt, and build on — with attribution.