Scientific infrastructure for Generative Search, GEO, governed evidence, metrics and reproducible AI research.
English · Español · Project status · Project matrix · Doctoral demo
Website · Research · Research Infrastructure · Scientific Dashboard · GitHub
GSLHub — Generative Search Lab Hub is an independent applied-research platform for studying how generative AI systems discover, select, cite, summarize and recommend digital information.
The platform is being developed as research infrastructure for the doctoral line:
From SEO to GEO (Generative Engine Optimization): development and validation of a scientific model to optimize organizational visibility in AI-based generative search engines.
GSLHub connects controlled experiments, prompt executions, observations, research artifacts, evidence, citations and governed metrics inside one auditable workflow.
Version 0.6.2 is the current Doctoral-ready product baseline.
It consolidates:
- responsive public frontend for mobile, tablet, laptop and desktop;
- responsive Payload administrator and scientific tables;
- custom Research Operations dashboard after CMS login;
- bilingual EN/ES Research Operations dashboard;
- light/dark theme support using Payload-native theme tokens;
- public bilingual Research Infrastructure demonstrator;
- direct private Research CMS access from the public frontend;
- public/private separation between dissemination and governed operations;
- deployment version-skew protection for Next.js assets;
- validated Hostinger cache-purge procedure after frontend redeploys.
No scientific schemas, metric calculators or governed research records were changed by the 0.6.x UI hotfixes.
GSLHub is designed around a simple scientific chain:
flowchart LR
A[Scientific problem] --> B[Hypothesis]
B --> C[Experiment]
C --> D[Execution]
D --> E[Evidence]
E --> F[Observation]
F --> G[Metrics]
F --> H[Citations]
G --> I[Reproducibility]
H --> I
I --> J[Public dissemination]
The complete operational matrix is preserved in docs/PROJECT-MATRIX.md so the system can be explained consistently in technical, scientific and doctoral contexts.
| Layer | Scientific purpose | Operational output |
|---|---|---|
| Scientific problem | Define what must be explained | Project / Benchmark scope |
| Hypothesis | State a testable expectation | Experiment hypothesis |
| Experiment | Define controlled method | Protocol, prompts, AI systems, repetitions |
| Execution | Run one governed trial | Prompt Execution snapshot |
| Evidence | Preserve the raw result | Research Artifact + Evidence |
| Observation | Code what was actually observed | Structured analytical record |
| Citations | Record source visibility | Source/domain and citation position |
| Metrics | Quantify outcomes | AIR, CR, MCP, RCR |
| Reproducibility | Prove integrity and repeatability | SHA-256, storage, recovery, lifecycle controls |
| Dissemination | Publish only safe outputs | Public dashboard / research pages |
The public Research Infrastructure demonstrator and the doctoral-demo runbook explain GSLHub through the same sequence:
Problem
→ Hypothesis
→ Experiment
→ Execution
→ Evidence
→ Metrics
→ Reproducibility
Public demonstrator:
- English:
https://gslhub.com/research-infrastructure - Español:
https://gslhub.com/es/research-infrastructure
Presentation runbook:
The public layer never needs to expose the full internal CMS schema or restricted research artifacts.
The final internal regression generated and verified a complete disposable research pipeline:
Full research pipeline TEST PASS
├── 5 Prompt Executions
├── 5 Observations
├── 5 Research Artifacts
├── 5 Evidence records
├── 3 Citations
└── 4 synthetic Metric records
Deterministic calculators
├── AIR = 3/4 = 0.75 PASS
├── CR = 2/4 = 0.50 PASS
├── MCP = 6/3 = 2.00 PASS
└── RCR = 3/4 = 0.75 PASS
TEST cleanup PASS
All disposable TEST batches and their generated records were removed successfully after validation.
The first complete development execution remains preserved as a non-doctoral validation record:
GSL-EXEC-GEO-0001 Completed / Published
├── GSL-ART-GEO-0001 Raw response / SHA-256 verified
├── GSL-ART-GEO-0002 Screenshot / SHA-256 verified
├── GSL-EVD-GEO-0001 Validated evidence
├── GSL-EVD-GEO-0002 Validated evidence
└── GSL-OBS-GEO-0001 Validated / Published observation
Reserved development executions remain untouched:
GSL-EXEC-GEO-0002 Planned
GSL-EXEC-GEO-0003 Planned
GSL-EXEC-GEO-0004 Planned
GSL-EXEC-GEO-0005 Planned
These are development-validation records, not doctoral findings.
| Code | Metric | Version | Unit |
|---|---|---|---|
| AIR | Answer Inclusion Rate | 0.1.0 | proportion |
| CR | Citation Rate | 0.1.0 | proportion |
| MCP | Mean Citation Position | 0.1.0 | position |
| RCR | Response Consistency Rate | 0.1.0 | proportion |
Metric calculators enforce eligibility and provenance rules. Target-specific metrics are not created when the required target or evidence does not exist.
GSLHub currently supports:
- versioned prompts, experiments and metric definitions;
- controlled repeated executions;
- immutable scientific snapshots after governed lifecycle transitions;
- direct Evidence ↔ Research Artifact provenance;
- persistent research-artifact storage outside deployment releases;
- SHA-256 integrity verification;
- quality-control and independent-review workflows;
- Development / Doctoral Research separation;
- controlled Final Development Reset;
- permanent storage-verification audits;
- documented restart, redeploy and recovery checks.
Authorized researchers use the private Payload CMS for governed operations. The Research Operations dashboard provides a presentation-friendly entry point while keeping full operational depth available through the underlying collections.
Primary operational areas:
- Research Environment;
- Experiments and Prompts;
- AI Systems;
- Prompt Executions;
- Observations;
- Research Artifacts;
- Evidence;
- Citations;
- Metric Definitions;
- Metrics;
- Storage Verifications.
The dashboard follows the selected Payload locale and the active Payload light/dark theme.
Production research artifacts are stored outside the Node.js deployment tree:
/home/<hostinger-user>/domains/gslhub.com/gslhub-data/research-artifacts
Verified sequence:
Upload → HTTP 200
Restart → same SHA-256 / HTTP 200
Redeploy → same SHA-256 / HTTP 200
Recovery drill → restored file / same SHA-256
GSLHub uses Next.js deployment versioning to reduce stale asset/version skew. Hostinger may additionally cache document responses outside the Node.js process.
Operational rule after frontend/CSS changes:
Deploy main
→ build/restart succeeds
→ purge Hostinger server cache
→ purge Hostinger CDN cache when enabled
→ desktop smoke test
→ mobile smoke test
This procedure prevents stale cached HTML from referencing assets from a previous deployment.
Research Environment Development Mode
Final Development Reset Not executed
Doctoral Research Mode Not activated
Real doctoral data 0
GSLHub remains in Development Mode until the doctoral protocol is frozen and the Final Development Reset produces a clean baseline.
The immediate focus is academic preparation rather than additional product features:
- prepare the doctoral proposal and research dossier;
- prepare the research-oriented CV;
- use GSLHub as the working demonstrator for thesis-supervisor discussions;
- freeze research questions, hypotheses, target dictionary, prompts, AI-system profiles and codebooks;
- preview and execute Final Development Reset;
- verify a clean baseline;
- activate Doctoral Research Mode;
- begin real doctoral data collection.
GSLHub platform 0.6.2
Payload CMS 3.75.0
Next.js 16.2.10
React 19.2.7
MongoDB driver 6.21.0
Database MongoDB Atlas
Hosting Hostinger
Artifact storage Persistent local storage outside deployment releases
Framework versions remain intentionally pinned until upgrades pass the complete administrator and scientific workflow on an isolated branch.
git clone https://github.com/gslhub/website.git
cd website
npm install
cp .env.example .env.local
npm run devQuality gate:
npm run lint
npm run typecheck
npm run build- Project matrix and research architecture
- Five-minute doctoral / supervisor demonstration
- Current operational project status — Spanish
- Spanish user manual
- First pilot protocol — Spanish
- Observation and citation codebook — Spanish
- Storage, backup and recovery procedure — Spanish
- English changelog
- Spanish changelog
The GSLHub software in this repository is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0-only). See LICENSE for the full license and NOTICE.md for copyright, third-party and brand information.
The software license does not grant trademark rights to the GSLHub name or associated brand identifiers. Research outputs or third-party materials may carry separate terms where explicitly indicated.
Copyright © 2026 Eduardo Yauri.
- Website: gslhub.com
- GitHub: github.com/gslhub
- Research email: research@gslhub.com
Research · GEO · Evidence · Metrics · Reproducibility · Open Science
Last updated: 18 August 2026