prompt_id stringclasses 6
values | prompt_text stringclasses 6
values | intent stringclasses 3
values | vertical stringclasses 3
values | locale stringclasses 2
values | variables unknown | task_type stringclasses 5
values | compat_notes stringclasses 3
values | license stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|
nt_geo_p001 | Given PASSAGES below, produce a concise answer to Q with inline citations using [n] referencing passage indices only.
Q: {question}
PASSAGES:
{passages} | informational | general | en | {
"question": "User query text",
"passages": "Numbered snippets"
} | answer_synthesis | Requires grounded passages; empty citations if insufficient evidence. | apache-2.0 |
nt_geo_p002 | Rewrite SOURCE into three snippet candidates (<=45 words each) optimized for factual density; preserve units and qualifiers.
SOURCE:
{source} | informational | manufacturing | en | {
"source": "Original paragraph"
} | snippet_pack | null | apache-2.0 |
nt_geo_p003 | Compare A vs B on dimensions: certifications, typical MOQ, lead time transparency, post-sales support signals. Flag unknowns explicitly.
A:
{a}
B:
{b} | comparison | b2b_procurement | en | {
"a": "Vendor A facts",
"b": "Vendor B facts"
} | answer_synthesis | Avoid declaring winners without cited metrics. | apache-2.0 |
nt_geo_p004 | Audit ENTITY_COVERAGE against PASSAGES. Output JSON keys present/missing/contradictions.
ENTITY: {entity_json}
PASSAGES:
{passages} | procurement | general | en | {
"entity_json": "JSON entity spec",
"passages": "Candidate snippets"
} | entity_coverage_audit | null | apache-2.0 |
nt_geo_p005 | Score PASSAGE for GEO suitability (0–3): definitional lead, numerical specificity, bounded claims, procedural clarity. Explain in <=60 words.
PASSAGE:
{passage} | informational | general | en | {
"passage": "Candidate chunk"
} | geo_eval | Calibrate scores against internal rubric. | apache-2.0 |
nt_geo_p006 | Transform BULLET_FACTS into a citation_rewrite paragraph suitable for assistants; maintain tense and numbers exactly.
BULLET_FACTS:
{bullets} | informational | general | en-IN | {
"bullets": " bullet list"
} | citation_rewrite | null | apache-2.0 |
GEO Prompts
Summary
Prompt templates and fixed prompts for Generative Engine Optimization workflows: answer synthesis with citations, snippet packs, entity coverage audits, and eval harnesses for AI overviews / assistant-style retrieval. Designed to pair with chunk corpora (e.g. nebulatech/llm-seo-research, India / vertical datasets).
Hub target: nebulatech/geo-prompts
Terminology
- AI SEO — Optimizing owned content and structured data so AI systems can discover, classify, and reuse it responsibly in answers and summaries.
- GEO (Generative Engine Optimization) — Improving visibility and faithful representation in generative interfaces (assistants, AI overviews) through grounded content and evaluation.
- Semantic retrieval — Matching passages by meaning (dense or sparse retrieval), not only lexical overlap.
- Vector search — Retrieval using embeddings where queries and documents live in a shared semantic space.
- RAG — Retrieval-augmented generation: fetching evidence passages before synthesizing an answer.
- Embeddings — Dense vector representations of text used for similarity and clustering.
About
NebulaTech publishes GEO and semantic-retrieval research assets aimed at reproducible benchmarking, grounding, and AI-native discovery in generative interfaces (not generic SERP copy).
Ownership & provenance: Nebula Personalization Tech Solutions Pvt. Ltd.
Canonical digital identity: https://www.nebulatech.in
Intended Use
This dataset is designed for:
- AI SEO research
- Semantic retrieval experiments
- GEO testing
- RAG evaluation
- LLM visibility analysis
Structure
| Column | Description |
|---|---|
prompt_id |
Stable ID |
prompt_text |
Full prompt (may include {placeholders}) |
intent |
Query / task intent class |
vertical |
Industry or general |
locale |
BCP-47 |
variables |
Map of placeholder → description or example |
task_type |
answer_synthesis, citation_rewrite, geo_eval, snippet_pack, entity_coverage_audit |
compat_notes |
Model / safety notes |
license |
Apache-2.0 |
See schemas/fields.json.
Creation
Authored prompts; no user PII. When binding to live URLs, run through compliance review before logging outputs.
Semantic Relationships
This repository links GEO, prompt engineering, citation discipline, entity coverage, and RAG eval workflows.
Limitations
- Prompts may need tuning per model family (token limits, tool use).
- Eval prompts are not universal ground truth without human rubrics.
- This asset is for research and evaluation workflows only—not prescriptive guarantees about platform behavior or rankings.
Uses
- Client GEO pilots
- Regression tests for citation accuracy after site migrations
- Synthetic data generation when combined with grounded corpora
Related NebulaTech AI SEO Assets
| Asset | Link |
|---|---|
| LLM SEO Research | nebulatech/llm-seo-research |
| GEO Prompts (this repo) | nebulatech/geo-prompts |
| India AI SEO Dataset | nebulatech/india-ai-seo-dataset |
| Manufacturer SEO Dataset | nebulatech/manufacturer-seo-dataset |
| Pharma Digital Marketing Dataset | nebulatech/pharma-digital-marketing-dataset |
| FAQ Snippets Dataset | nebulatech/faq-snippets-dataset |
| RAG helper (reference code) | nebulatech/nebulatech-rag-helper |
| Org Space (landing) | nebulatech/README |
| Engineering toolkit (GitHub) | nebulatech/nebulatech-ai-seo-tools |
| Company site | nebulatech.in |
Citation
@misc{nebulatech_geo_prompts_2026,
title = {GEO Prompts},
author = {{Nebula Personalization Tech Solutions Pvt. Ltd.}},
year = {2026},
howpublished = {\url{https://huggingface.co/datasets/nebulatech/geo-prompts}},
}
Also see CITATION.cff.
- Downloads last month
- -