CRLPCache Refresh & Lineage Protocol
E.I.A.R.(V)-AERiV · VEHICLE Systems Lab

Verification
before reasoning.

CRLP is an experimental source-integrity protocol for AI systems, agents and audit workflows. It helps verify whether a source is current, coherent and safe to reason from before an AI cites it, summarizes it, audits it or acts on it.

Download CRLP to detect stale cache, broken previews, version drift, ambiguous publication states and mismatched public builds before they contaminate AI reasoning.
Download CRLP Experimental View Real Experiments
Experimental protocol · Source integrity

CRLP Experimental v0.1

CRLP does not attack caches. It observes cache state, compares versions, checks markers and hashes, separates preview failure from file loss, and classifies whether a source is safe to reason from before AI systems use it for retrieval, citation, audit or decision-support.

What CRLP helps resolve
Stale-while-revalidate responses that still look valid to an AI pipeline.
CDN HIT layers that keep serving inherited content after an origin change.
Broken preview layers that hide the fact that direct downloads still exist.
Publication states that appear incomplete until repository visibility is verified.
Alias versus deployment mismatches across public domains and direct builds.
Header ambiguity when Cache-Control, Age or ETag are not enough by themselves.
Understand the protocol in one pass
01 SourceOpen

Verify the URL, file, record or repository identity first.

02 CacheCompare

Read the source normally and through a probe path to surface divergence.

03 WitnessHash + Marker

Check markers, hashes, dates and lineage before trusting availability.

04 DecisionAuthorize

Classify GREEN, YELLOW, ORANGE, RED or BLACK before reasoning.

Protocol promiseCRLP verifies whether a source is safe to reason from.
It does not purge caches, bypass infrastructure or claim to fix every cache problem.
Download CRLP Experimental
Experimental release · File coming soon
Attribution required:
VEHICLE Systems Lab · Roberto Borda Milan
Research collaboration →
Why CRLP exists
The Problem

AI can reason over stale or incomplete sources.

Modern AI systems, RAG pipelines and web agents may read sources through cached pages, stale CDN responses, outdated indexes, broken previews, draft-only publication states or ambiguous headers. A source may exist and still be unsafe to use.

The Method

Identity, temporal state, cache, lineage and authorization.

CRLP applies a layered verification structure derived from the Borda Milan Pyramid. A source is not accepted only because it exists; it must pass identity, temporal, cache, lineage, version and authorization checks.

Real Experiments

Live observations already conducted on public infrastructure.

  • LIVE-TC-001 / LIVE-TC-002: Vercel cache and controlled marker verification.
  • LIVE-TC-ZENODO-001: public preprint visibility with DOI, PDF, DOCX and ZIP confirmation.
  • LIVE-TC-ZENODO-003: preview failure separated from real download availability.
Decisions

Use the source, warn, re-check or block.

CRLP classifies source integrity with GREEN, YELLOW, ORANGE, RED and BLACK decisions. That gives AI developers and auditors a clear threshold before a source enters reasoning, citation or decision-support.

License & Terms

CC BY 4.0, experimental, and meant to be tested responsibly.

CRLP v0.1 Experimental is released under the Creative Commons Attribution 4.0 International License (CC BY 4.0). You are free to share, adapt, integrate and build upon this work for any purpose, including commercial use, provided that appropriate attribution to VEHICLE Systems Lab and Roberto Borda Milan is preserved.

Suggested attribution: “CRLP — Cache Refresh & Lineage Protocol, developed by VEHICLE Systems Lab / Roberto Borda Milan.” This package is experimental and evolving. VEHICLE Systems Lab does not present CRLP as an official AI safety standard, infrastructure security standard or regulatory compliance framework.

Claims about effectiveness should be tested, documented and validated through authorized use. CRLP does not purge, attack, bypass or modify third-party infrastructure. It is an observation, verification and source-integrity protocol. Full license text: Creative Commons Attribution 4.0.

How to Use CRLP

Local experimental use with PowerShell.

CRLP v0.1 Experimental can be tested locally using read-only probes. These probes do not purge cache, modify production, bypass authorization or attack infrastructure. They observe public or authorized sources and generate JSON reports.

  • Step 1. Download the probe script, for example crlp_live_fetch_probe_v0_2.py.
  • Step 2. Open PowerShell in the working folder: cd "C:\Users\YOUR_USER\Downloads\YOUR_PROJECT_FOLDER"
  • Step 3. Run a basic check: python .\crlp_live_fetch_probe_v0_2.py https://example.com --out crlp_probe_report.json --sleep 2 --marker "EXPECTED_MARKER" --save-body --body-prefix crlp_body
  • Step 4. Review the JSON report: type .\crlp_probe_report.json
  • Step 5. Interpret the decision: GREEN, YELLOW, ORANGE, RED or BLACK.

CRLP does not mean “clean every cache.” CRLP means “verify whether the source is safe to reason from.”

Use Models

Free protocol, professional integration and advanced AI-ready deployments.

  • 1. Free Experimental Version. Best for local testing, research validation and first experiments. Includes read-only probes for cache state, content hashes, markers, preview/download integrity, DOI visibility and publication-state coherence.
  • 2. CRLP CPU Server for Companies. Designed for repositories, RAG systems, documentation platforms and AI teams that need repeatable source verification without requiring GPU for the first operational layer.
  • 3. Advanced GPU Server / AI-Agent Integration. For semantic comparison between versions, embedding-based similarity analysis, large-scale corpus auditing, advanced lineage reconstruction and specialized AI agents.

For institutional integration, CPU server deployment, advanced GPU architecture, custom AI-agent integration or technical support, contact contact@vehiclesystemslab.com.

Who Needs CRLP?

Any organization that cannot afford stale, incomplete or misleading sources.

  • Generative AI companies.
  • RAG platforms and enterprise search systems.
  • Cloud, CDN and hosting platforms.
  • Scientific and academic repositories.
  • Legal, financial and medical AI systems.
  • Cybersecurity and compliance teams.
  • Governments, archives and public documentation platforms.
  • Technical documentation companies.
Professional Integration

Free to test. Professional to integrate.

The free version is intended for open testing and local adoption. Professional deployment is where CRLP becomes operational infrastructure: internal source-integrity layers, documented audit workflows, company-specific integration patterns, CPU server deployments, GPU-supported evidence analysis and technical coordination from VEHICLE Systems Lab.

For research collaboration, institutional integration and custom architecture, contact contact@vehiclesystemslab.com.

Other VEHICLE projects
VEHICLE ACCESS →VEHICLE CPS →VEHICLE ODI →VEHICLE SUPRA →← Back to MADRE