34 Beginner Friendly – Raw Semantic Discovery GPT Prompts for MIRENA

Use Raw Semantic Discovery prompts when MIRENA needs to collect semantic evidence before building a topical map, content brief, rewrite plan, entity layer, internal link plan, information gain pass, SERP feature plan, or schema notes.

This is the upstream discovery layer.

It collects candidate entities, concepts, modifiers, query paths, intent signals, SERP patterns, competitor signals, source signals, and opportunity notes.

Start with source context.

Do not run raw discovery against a loose keyword file, sitemap, competitor list, analytics export, behavior report, or draft until MIRENA knows the site, offer, audience, allowed topics, blocked topics, internal link rules, and next workflow stage.

Use the source context template before this workflow if the project base is not ready. Use Getting Started with MIRENA if you need the full onboarding route before running a discovery workflow.

Start with Source Context Before Discovery

Source context controls the discovery layer.

It tells MIRENA what the site is, who it serves, what it sells, which topics belong, which topics are blocked, and how the output should move into the next workflow.

Raw discovery can collect too much. A keyword export, crawl, sitemap, competitor set, or SERP review can contain topics that look related but do not support the site.

The Source Context page explains the rule behind this: source context protects the site’s topical focus before new pages enter the map.

Use source context to define:

• site purpose
• audience
• offer
• region
• allowed topics
• blocked topics
• existing pages
• protected pages
• internal link targets
• commercial routes
• workflow goal
• output format
• next workflow stage

When source context is missing, stop and build it first.

What Raw Semantic Discovery Does

Raw Semantic Discovery collects the semantic evidence that later MIRENA workflows use.

It does not decide the finished site structure.

It does not write the content brief.

It does not rewrite the page.

It does not finalize schema.

It gathers, labels, filters, and routes the evidence first.

The output can feed Topical Maps + Planning when the next job is structure. It can feed Content Briefs when the next job is writer instruction. It can feed Drafting + Rewriting when the next job is page repair. It can also feed information gain work when the next job is useful differentiation.

Use MIRENA outputs when you need to define what the discovery package should return.

Raw discovery is useful when you have files, exports, pages, or competitor sources, but not yet a clean structure.

You may have a keyword export, sitemap, list of competitor URLs, content crawl, old briefs, GSC queries, GA4 landing page data, or behavior notes. Raw discovery helps MIRENA turn those inputs into candidate signals before another workflow decides what to keep, reject, merge, map, brief, rewrite, or route.

What This Page Does Not Repeat

This page does not repeat the Entity SEO and Salience prompt set.

The Entity SEO and Salience guide covers entity maps, entity definitions, entity salience, entity placement, support pages, entity consistency, entity gaps, entity relationships, and entity markup cues.

This page comes before that.

It gives MIRENA the raw discovery evidence that later workflows can organize, score, repair, map, brief, or prepare for schema.

Use this page when the job is collection, extraction, classification, expansion, benchmarking, or handoff.

Use Entity SEO when the next job is entity structure, entity meaning, entity relationships, salience, or markup planning.

When to Use Raw Semantic Discovery Prompts

Use this prompt collection when you need to discover what is inside the data before planning the next step.

Raw Semantic Discovery is useful for:

• new project intake
• keyword export analysis
• GSC query discovery
• sitemap review
• content crawl review
• competitor URL review
• SERP pattern intake
• content refresh research
• topical map preparation
• content brief preparation
• rewrite intake
• information gain preparation
• internal link planning intake
• schema cue preparation after page approval

Run Raw Semantic Discovery before building a processed topical map when the source inputs are messy.

Run it before briefing or rewriting when the page target is unclear, the topic has many modifiers, or the SERP has mixed intent.

Use MIRENA inputs when you need to decide which files should feed the discovery pass. Use the MIRENA workflow when the discovery package needs a clear route into mapping, briefing, rewriting, linking, or schema notes.

The Raw Discovery Workflow

Run the Raw Semantic Discovery workflow in this order.

1. Check source context.
2. Review discovery assets.
3. Scan the corpus.
4. Extract candidates.
5. Classify candidate types.
6. Weight candidates.
7. Harvest modifiers.
8. Harvest candidate attributes.
9. Classify query intent.
10. Select query treatments.
11. Generate synthetic queries.
12. Classify synthetic queries.
13. Cluster semantic queries.
14. Find latent intents.
15. Expand the query network.
16. Harvest SERP entities.
17. Capture SERP patterns.
18. Harvest competitor candidates.
19. Snapshot competitive coverage.
20. Scan SERP consensus.
21. Scan SERP divergence.
22. Build a topical authority baseline.
23. Scan competitor schema.
24. Expand semantic neighborhoods.
25. Rank source candidates.
26. Build the opportunity matrix.
27. Package the entity universe.
28. Discover page archetype seeds.
29. Hand off the discovery output.

Do not move into topical mapping, briefs, rewrites, internal links, or schema if the raw discovery package contains unclear, off-scope, or unsupported candidates.

The Raw Discovery Prompt Pattern

Use short commands when the task is clear.

Use expanded prompts when you need MIRENA to control scope, fields, exclusions, output format, and routing.

Short command pattern:

text

Run [raw discovery module] on [asset].

Expanded prompt pattern:

text

Run [raw discovery module] on [asset].

Use the source context first.

Analyze [asset] for [specific discovery goal].

Return the output with these fields:
- field 1
- field 2
- field 3
- field 4
- field 5
- next workflow route

Reject anything that does not fit the source context.

Flag uncertain candidates for review.

Route the final output into [next workflow stage].

A short prompt is enough for simple tasks.

An expanded prompt is better when the input is large, messy, risky, or ready for handoff.

What to Give MIRENA Before Running Raw Discovery

Start with source context.

Then add the strongest evidence available.

For a keyword file, give MIRENA:

• source context
• keyword export
• target topic
• target audience
• region
• commercial goal
• known page targets
• current sitemap if available

For a sitemap or crawl, give MIRENA:

• source context
• sitemap or URL list
• page titles
• meta titles if available
• page roles if known
• internal link data if available
• pages to protect
• pages already flagged for refresh

For a competitor review, give MIRENA:

• source context
• competitor URLs
• target query or query group
• SERP notes if available
• target market
• pages already published on your site
• output you want after discovery

For search and analytics files, give MIRENA:

• source context
• GSC pages export
• GSC queries export
• GA4 landing pages export
• GA4 engagement data
• date range
• known conversion goals
• pages that need review

For behavior data, give MIRENA:

• source context
• heatmap notes
• scroll depth notes
• click tracking notes
• session recording notes
• internal site search logs
• form tracking notes
• page or cluster target

Use Semantic SEO when the discovery task needs stronger meaning, context, and topic fit before page planning.

Raw Semantic Discovery Modules

The modules below collect raw semantic evidence without repeating the Entity SEO and Salience prompt set.

Choose the smallest module that fits the job.

1. Source Context Check

Use this before any raw discovery module.

Short command:

text

Run Source Context Check for this discovery task.

Expanded prompt:

text

Run Source Context Check for this discovery task.

Use the source context first.

Review the project boundary before any extraction, clustering, competitor review, or query expansion starts.

Return the output with these fields:
- site purpose
- audience
- offer
- allowed topic lanes
- blocked topic lanes
- target workflow stage
- discovery boundary
- files that should influence discovery
- files that should not influence discovery
- risks if discovery starts too broad
- next workflow route

Reject topics, candidates, or query paths that do not support the source context.

Flag any missing source context needed before discovery can continue.

Route the final output into Discovery Asset Review or Corpus Scan.

Best for:

• new projects
• imported keyword files
• unclear page lists
• broad topics
• competitor research

Output should include:

• allowed topic lanes
• blocked topic lanes
• audience notes
• offer notes
• discovery boundary
• next stage

Use this to:

Keep the discovery layer tied to the site. Source context stops MIRENA from collecting candidates and query paths that do not belong on the project.

If the source context is weak, stop here and build it before running the next discovery module.

2. Discovery Asset Review

Use this when the project has multiple files or data sources.

Short command:

text

Run Discovery Asset Review on these files.

Expanded prompt:

text

Run Discovery Asset Review on these files.

Use the source context first.

Review each uploaded file, export, URL list, report, brief, sitemap, crawl, search file, analytics file, behavior note, competitor source, or previous MIRENA output.

Do not extract candidates yet.

Return the output with these fields:
- asset name or label
- asset type
- source quality
- freshness risk
- discovery value
- best use
- noise risk
- missing file or missing field
- recommended processing order
- keep, hold, ignore, or review
- next workflow route

Separate assets that should feed raw discovery from assets that should be held for later workflows.

Flag any file that may create topic drift or weak candidates.

Route the final output into Corpus Scan, NER Pass, Query Intent Classification, SERP Entity Harvest, or Source Candidate Ranking.

Best for:

• source context files
• sitemap exports
• keyword exports
• SERP exports
• GSC files
• GA4 files
• competitor URLs
• crawl data

Output should include:

• asset type
• asset quality
• best use
• risk notes
• missing files
• recommended discovery order

Use this to:

Separate useful discovery files from files that should be held for later. Some files help with raw evidence. Others are better for rewrites, internal links, or schema review.

Discovery Asset Review helps MIRENA decide which files should influence the first pass.

3. Corpus Scan

Use this to scan a page set, draft set, or content export before extraction.

Short command:

text

Run Corpus Scan on this content set.

Expanded prompt:

text

Run Corpus Scan on this content set.

Use the source context first.

Scan the content before running deeper extraction.

Look for recurring terms, recurring concepts, repeated phrases, page themes, weak signals, overused language, underdeveloped concepts, candidate entities, modifier candidates, and content sections that need deeper review.

Return the output with these fields:
- source page or document
- recurring term
- recurring concept
- repeated phrase
- candidate entity
- modifier candidate
- weak signal
- overused signal
- source location
- extraction confidence
- suggested next module
- next workflow route

Do not turn the output into a topical map.

Do not build entity structure yet.

Route the output into NER Pass, Concept Harvest, Frequency Signal Scan, Placement Signal Scan, or Entity Universe Package.

Best for:

• existing website content
• old blog libraries
• content refresh projects
• draft folders
• docs sections
• page inventories

Output should include:

• recurring concepts
• repeated phrases
• entity candidates
• modifier candidates
• low confidence terms
• extraction notes

Use this to:

Give MIRENA the first read of a content set. It finds repeated patterns before deeper extraction begins.

Run this before NER Pass when you have a large page set or mixed content export.

4. NER Pass

Use this to extract named entities from text, page sets, or competitor content.

Short command:

text

Run NER Pass on this asset.

Expanded prompt:

text

Run NER Pass on this asset.

Use the source context first.

Extract named entity candidates from the asset.

Include people, organizations, products, software, brands, places, known concepts, named frameworks, product features, documents, standards, tools, and repeated named references.

Return the output with these fields:
- candidate entity
- entity type
- source page or source passage
- source location
- frequency signal
- placement signal
- confidence level
- source context fit
- keep, reject, or review
- reason
- suggested next module
- next workflow route

Do not build an entity map.

Do not score final salience.

Do not create schema.

Keep this as a discovery extraction pass.

Route the output into Entity Type Classification, Candidate Weighting, Entity Universe Package, or downstream entity review.

Best for:

• pages
• drafts
• SERP exports
• competitor pages
• product docs
• source files

Output should include:

• candidate entity
• entity type
• source location
• frequency signal
• placement signal
• confidence level

Use this to:

Extract candidate entities before MIRENA decides which ones are worth keeping.

This is a discovery action, not the finished entity model. Later workflows can organize, rank, repair, or apply the candidates.

5. Concept Harvest

Use this when the asset contains important non-named concepts.

Short command:

text

Run Concept Harvest on this asset.

Expanded prompt:

text

Run Concept Harvest on this asset.

Use the source context first.

Extract important non-named concepts that may not appear as named entities.

Look for process terms, workflow terms, category terms, decision concepts, scoring ideas, user states, intent concepts, product concepts, technical ideas, and repeated support concepts.

Return the output with these fields:
- concept
- concept type
- source passage
- source location
- related topic
- related candidate entity
- confidence note
- source context fit
- keep, reject, or review
- reason
- suggested next module
- next workflow route

Do not turn the concepts into a map yet.

Do not assign final entity roles.

Route the output into Candidate Weighting, Semantic Neighborhood Expansion, Discovery Opportunity Matrix, or Entity Universe Package.

Best for:

• semantic SEO pages
• guides
• documentation
• workflow pages
• educational clusters
• query sets

Output should include:

• concept name
• concept type
• source passage
• related topic
• confidence note
• next stage

Use this to:

Capture important process terms, decision rules, scoring concepts, and user state language before mapping or briefing.

Some important semantic signals are not named entities. Concept Harvest keeps those signals from being missed.

6. Entity Type Classification

Use this when raw extraction returns mixed candidates.

Short command:

text

Run Entity Type Classification on this candidate list.

Expanded prompt:

text

Run Entity Type Classification on this candidate list.

Use the source context first.

Normalize the raw candidate list into clear candidate types.

Classify each candidate as person, organization, brand, software, product, feature, location, process, framework, document, category, metric, query modifier, content format, support concept, or reject.

Return the output with these fields:
- candidate
- normalized type
- source context fit
- confidence
- unclear type warning
- reject reason
- preferred label
- duplicate label if found
- suggested next module
- next workflow route

Do not decide final hierarchy.

Do not run entity salience.

Do not create schema cues.

Route the output into Candidate Weighting, Entity Universe Package, or Discovery Opportunity Matrix.

Best for:

• NER outputs
• competitor extracts
• product lists
• mixed keyword files
• SERP entity harvests

Output should include:

• candidate entity
• normalized type
• confidence
• unclear type warning
• reject reason
• review note

Use this to:

Clean the candidate layer before it moves downstream.

Raw extraction often mixes people, tools, brands, topics, categories, locations, features, and vague phrases. This module helps MIRENA sort them before weighting or packaging.

7. Candidate Weighting

Use this before an entity universe is built.

Short command:

text

Run Candidate Weighting on this discovery list.

Expanded prompt:

text

Run Candidate Weighting on this discovery list.

Use the source context first.

Weight each candidate by frequency, placement, source quality, source count, query intent fit, topic fit, buyer fit, workflow fit, and downstream usefulness.

Return the output with these fields:
- candidate
- candidate type
- frequency score
- placement score
- source quality score
- source count
- intent fit
- topic fit
- buyer fit
- workflow fit
- risk note
- keep, reject, or review
- reason
- next workflow route

Reject candidates that are repeated but off-scope.

Flag candidates that look useful but need more evidence.

Route the output into Entity Universe Package, Discovery Opportunity Matrix, or Topical Maps and Planning.

Best for:

• large extraction sets
• raw entity harvests
• keyword to concept work
• competitor scans
• SERP exports

Output should include:

• candidate
• frequency score
• placement score
• source quality score
• intent fit
• topic fit
• keep or reject note

Use this to:

Separate useful signals from noise before the output moves into a map, brief, or rewrite.

A repeated candidate is not always useful. Candidate Weighting gives MIRENA a reason to keep, reject, or review each item.

8. Frequency Signal Scan

Use this when frequency may show repeated semantic patterns.

Short command:

text

Run Frequency Signal Scan on this content set.

Expanded prompt:

text

Run Frequency Signal Scan on this content set.

Use the source context first.

Scan the content set for repeated terms, repeated concepts, recurring phrases, repeated modifiers, repeated entity candidates, and repeated page themes.

Return the output with these fields:
- term or concept
- frequency
- source count
- dominant source
- repeated context
- topic fit
- possible overuse
- possible importance
- risk note
- suggested next module
- next workflow route

Do not treat frequency as proof of importance by itself.

Flag high-frequency off-scope candidates.

Flag repeated terms that may indicate drift, duplication, or weak copy.

Route the output into Candidate Weighting, Placement Signal Scan, Corpus Scan Review, or Discovery Opportunity Matrix.

Best for:

• old content libraries
• competitor SERP sets
• query exports
• page clusters
• draft collections

Output should include:

• term or concept
• frequency
• source count
• dominant source
• risk note
• next step

Use this to:

Find what a page set repeatedly emphasizes.

Frequency can also reveal repetition, overuse, and topic drift. Use this module to find patterns before deciding structure.

9. Placement Signal Scan

Use this when you need to know where raw candidates appear.

Short command:

text

Run Placement Signal Scan on this asset.

Expanded prompt:

text

Run Placement Signal Scan on this asset.

Use the source context first.

Check where raw candidates appear across titles, meta titles, headings, openings, body sections, tables, FAQs, captions, internal anchors, schema notes, and CTA sections.

Return the output with these fields:
- candidate
- source page
- placement location
- placement strength
- repeated placement pattern
- source context fit
- importance note
- extraction confidence
- suggested next module
- next workflow route

Do not score final salience.

Do not rewrite the page.

Use placement only as a discovery signal.

Route the output into Candidate Weighting, SERP Pattern Intake, Content Briefs, or Drafting and Rewriting.

Best for:

• competitor page analysis
• draft analysis
• snippet analysis
• content refresh
• SERP structure review

Output should include:

• candidate
• placement location
• source page
• importance note
• extraction confidence
• next stage

Use this to:

Understand where concepts appear before assigning importance.

A candidate that appears in titles and headings has a stronger placement signal than one buried deep in a paragraph.

10. Modifier Harvest

Use this when raw queries or pages contain useful modifiers.

Short command:

text

Run Modifier Harvest on this keyword set.

Expanded prompt:

text

Run Modifier Harvest on this keyword set.

Use the source context first.

Extract modifiers that change audience, product fit, feature angle, location, comparison need, process stage, price sensitivity, problem state, buyer stage, content format, or page type.

Return the output with these fields:
- modifier
- modifier type
- example query
- related candidate
- intent signal
- page type signal
- source context fit
- keep, reject, or review
- reason
- suggested next module
- next workflow route

Group modifiers by type.

Flag modifiers that point to topics outside the source context.

Route the output into Facet Intent Extraction, Query Modifier Scan, Query Treatment Selection, or Topical Maps and Planning.

Best for:

• keyword exports
• GSC query exports
• SERP exports
• local SEO projects
• product page planning
• comparison clusters

Output should include:

• modifier
• modifier type
• related candidate
• intent signal
• page type suggestion
• next stage

Use this to:

Show how users frame a topic.

A modifier can reveal audience, location, product type, comparison need, problem state, price sensitivity, or process stage.

11. Attribute Candidate Harvest

Use this before a later workflow assigns attributes.

Short command:

text

Run Attribute Candidate Harvest on this corpus.

Expanded prompt:

text

Run Attribute Candidate Harvest on this corpus.

Use the source context first.

Collect candidate attributes that may define, describe, compare, qualify, or clarify important candidates.

Look for feature terms, qualities, constraints, use cases, benefits, limitations, categories, criteria, specifications, proof points, and repeated descriptive phrases.

Return the output with these fields:
- candidate attribute
- related candidate
- source passage
- source location
- modifier link
- confidence level
- source context fit
- keep, reject, or review
- reason
- suggested downstream workflow
- handoff note

Do not assign final attribute priority.

Do not repeat the later Entity Attributes workflow.

This module only collects candidate attributes for downstream review.

Route the output into Entity Universe Package, Content Briefs, Information Gain, or Entity SEO and Salience.

Best for:

• product pages
• comparison pages
• documentation
• competitor pages
• query exports

Output should include:

• candidate attribute
• source passage
• linked candidate entity
• modifier link
• confidence
• handoff note

Use this to:

Collect attribute candidates without deciding their final priority.

The later entity workflow can decide which attributes define the entity, which ones belong in the page, and which ones should be removed.

12. Facet Intent Extraction

Use this when modifiers imply different user needs.

Short command:

text

Run Facet Intent Extraction on this query set.

Expanded prompt:

text

Run Facet Intent Extraction on this query set.

Use the source context first.

Identify the facets that change user need or page treatment.

Extract feature facets, audience facets, location facets, comparison facets, price facets, problem facets, process facets, trust facets, format facets, and urgency facets.

Return the output with these fields:
- facet
- facet type
- related query
- primary intent label
- secondary intent label
- confidence score
- page treatment suggestion
- source context fit
- keep, reject, or review
- reason
- next workflow route

Do not build the page map yet.

Use facets to prepare cleaner query treatment and page planning.

Route the output into Query Intent Classification, Query Treatment Selection, or Topical Maps and Planning.

Best for:

• long tail queries
• product attributes
• local modifiers
• comparison modifiers
• feature modifiers

Output should include:

• facet
• related query
• intent label
• confidence score
• page treatment
• review note

Use this to:

Show the angle of the search.

A query may be about the same topic but through a different feature, audience, location, comparison, or problem frame.

13. Query Intent Classification

Use this when a keyword export needs intent labels before mapping.

Short command:

text

Run Query Intent Classification on this query set.

Expanded prompt:

text

Run Query Intent Classification on this query set.

Use the source context first.

Classify each query by primary intent, secondary intent, user stage, likely page type, and content treatment.

Flag mixed intent, unclear intent, commercial investigation queries, informational queries, comparison queries, navigational queries, product-led queries, support queries, local queries, and queries that should not become pages.

Return the output with these fields:
- query
- primary intent
- secondary intent
- user stage
- page type recommendation
- content treatment
- SERP format note
- source context fit
- keep, merge, section, FAQ, comparison, anchor, or reject
- reason
- next workflow route

Do not convert the query list into a topical map yet.

Route the output into Topical Maps and Planning if the query group needs structure.

Route it into Content Briefs if the page target is already clear.

Route it into Information Gain if the SERP looks repetitive.

Best for:

• keyword exports
• GSC query exports
• synthetic query lists
• SERP query groups
• content brief intake

Output should include:

• query
• intent label
• secondary intent
• conflict warning
• content treatment
• SERP format note

Use this to:

Stop keyword data from becoming a page plan too early.

Query Intent Classification helps MIRENA decide which queries may need pages, sections, FAQs, comparisons, anchors, or rejection.

14. Query Modifier Scan

Use this when query wording signals different page needs.

Short command:

text

Run Query Modifier Scan on this query set.

Expanded prompt:

text

Run Query Modifier Scan on this query set.

Use the source context first.

Scan query wording for modifiers that change page type, intent, format, audience, product stage, local need, comparison need, trust need, or next step.

Return the output with these fields:
- modifier
- modifier class
- example query
- repeated pattern
- intent shift
- page type signal
- source context fit
- keep, reject, or review
- reason
- next workflow route

Group repeated modifiers.

Flag modifiers that could create unnecessary pages.

Flag modifiers that should become sections, FAQs, comparison blocks, anchors, templates, or examples.

Route the output into Query Treatment Selection, Semantic Query Clustering, or Topical Maps and Planning.

Best for:

• keyword exports
• search console data
• topic expansion
• page type decisions
• query gap work

Output should include:

• modifier
• modifier class
• example query
• intent shift
• page type signal
• next stage

Use this to:

Find the words that change the page job.

“Template,” “example,” “software,” “near me,” “best,” “vs,” and “how to” all point to different content treatments.

15. Query Treatment Selection

Use this when MIRENA must decide how a query should be handled.

Short command:

text

Run Query Treatment Selection on this query list.

Expanded prompt:

text

Run Query Treatment Selection on this query list.

Use the source context first.

Decide how each query should be handled before any new page is created.

Choose from:
- dedicated page
- section
- FAQ
- list
- comparison block
- table
- anchor text target
- internal link target
- template
- example
- merge into existing page
- reject

Return the output with these fields:
- query
- recommended treatment
- target page if known
- page type if new
- reason
- source context fit
- overlap risk
- internal link note
- keep, merge, section, FAQ, anchor, or reject
- next workflow route

Reject off-scope queries.

Flag queries that look like duplicate intent.

Route the output into Topical Maps and Planning, Content Briefs, Internal Linking, or Information Gain.

Best for:

• raw keyword lists
• synthetic queries
• FAQ planning
• page vs section prep
• topical map intake

Output should include:

• query
• recommended treatment
• reason
• source context fit
• risk note
• next stage

Use this to:

Decide if a query deserves a page, section, FAQ answer, anchor, comparison block, list, or rejection.

This module prepares query handling before the processed topical map is built.

16. Synthetic Query Generation

Use this when you want MIRENA to generate possible future or long tail queries.

Short command:

text

Run Synthetic Query Generation from this entity and modifier set.

Expanded prompt:

text

Run Synthetic Query Generation from this candidate and modifier set.

Use the source context first.

Generate inferred queries from accepted candidates, modifiers, user stages, page types, product angles, audience needs, location signals, feature signals, and comparison paths.

Return the output with these fields:
- synthetic query
- source candidate
- source modifier
- inferred intent
- likely user stage
- content treatment
- source context fit
- confidence
- risk note
- keep, reject, or review
- next workflow route

Do not treat generated queries as proven demand.

Flag speculative queries.

Reject generated queries that do not fit the source context.

Route the output into Synthetic Query Classification, Query Treatment Selection, or Semantic Query Clustering.

Best for:

• zero volume topics
• emerging topics
• long tail planning
• product attributes
• local modifiers
• audience variants

Output should include:

• synthetic query
• source entity
• source modifier
• inferred intent
• content treatment
• risk note

Use this to:

Explore possible search paths around emerging products, niche workflows, or new categories.

Not every useful query appears in a keyword export, but generated queries still need review before they enter a map.

17. Synthetic Query Classification

Use this after synthetic queries are generated.

Short command:

text

Run Synthetic Query Classification on this generated query list.

Expanded prompt:

text

Run Synthetic Query Classification on this generated query list.

Use the source context first.

Review each generated query for intent, confidence, source context fit, usefulness, risk, and downstream value.

Return the output with these fields:
- generated query
- source candidate
- source modifier
- primary intent
- secondary intent
- confidence
- usefulness
- risk
- keep, revise, reject, or review
- reason
- next workflow route

Reject generated queries that create topic drift.

Flag queries that need SERP validation.

Route the output into Query Treatment Selection, Semantic Query Clustering, SERP Pattern Intake, or Topical Maps and Planning.

Best for:

• generated query sets
• future demand planning
• speculative content planning
• audience-specific queries
• product feature expansions

Output should include:

• query
• intent
• context
• confidence
• keep, revise, or reject
• next stage

Use this to:

Prevent speculative ideas from entering the map without confidence checks.

Generated queries can be useful, but they should be classified before page planning begins.

18. Semantic Query Clustering

Use this when raw queries need meaningful groups before page decisions.

Short command:

text

Run Semantic Query Clustering on this query set.

Expanded prompt:

text

Run Semantic Query Clustering on this query set.

Use the source context first.

Group queries by meaning, intent layer, user job, page type, modifier pattern, and shared concept.

Do not group queries only by shared words.

Return the output with these fields:
- cluster name
- included queries
- shared concept
- dominant intent
- secondary intent
- likely page type
- page vs section risk
- overlap risk
- source context fit
- suggested next module
- next workflow route

Flag clusters that mix incompatible intents.

Flag clusters that may need separate pages.

Route the output into Query Treatment Selection, Topical Maps and Planning, Content Briefs, or SERP Pattern Intake.

Best for:

• keyword exports
• GSC exports
• synthetic query lists
• SERP query sets
• topical map intake

Output should include:

• query cluster
• shared concept
• dominant intent
• secondary intent
• cluster risk
• next stage

Use this to:

Group queries by meaning, not only by repeated words.

This helps MIRENA avoid weak keyword buckets and build better downstream page decisions.

19. Latent Intent Discovery

Use this when visible queries do not explain the full user need.

Short command:

text

Run Latent Intent Discovery on this topic.

Expanded prompt:

text

Run Latent Intent Discovery on this topic.

Use the source context first.

Find hidden user needs that are implied by the topic, query set, SERP, competitors, modifiers, product context, or user stage.

Look for emerging needs, comparison needs, proof needs, trust needs, support needs, workflow needs, local needs, and buyer confidence needs.

Return the output with these fields:
- latent intent
- source signal
- related query pattern
- user need
- likely page type
- content treatment
- source context fit
- confidence
- risk note
- next workflow route

Do not create new pages from latent intent alone.

Flag intents that need SERP validation, query evidence, or source context review.

Route the output into Query Network Expansion, SERP Pattern Intake, Topical Maps and Planning, or Content Briefs.

Best for:

• emerging topics
• low volume topics
• ambiguous topics
• new product features
• competitor gaps

Output should include:

• latent intent
• source signal
• query pattern
• user need
• content treatment
• risk note

Use this to:

Find user needs that keyword data may not show clearly.

Latent Intent Discovery helps MIRENA see hidden search paths before the map or brief is created.

20. Query Network Expansion

Use this when a topic needs a broader query network.

Short command:

text

Run Query Network Expansion on this seed topic.

Expanded prompt:

text

Run Query Network Expansion on this seed topic.

Use the source context first.

Expand the seed topic into related query paths, intent branches, user stages, support questions, comparison paths, feature paths, process paths, and adjacent topics.

Return the output with these fields:
- query path
- branch type
- related concept
- primary intent
- secondary intent
- likely page treatment
- source context fit
- keep, reject, or review
- reason
- suggested next workflow

Reject branches that drift outside the source context.

Flag branches that should become sections instead of pages.

Route the output into Semantic Query Clustering, Query Treatment Selection, Topical Maps and Planning, or Content Briefs.

Best for:

• early topic exploration
• topical map intake
• content brief planning
• FAQ discovery
• search journey mapping

Output should include:

• query path
• branch type
• intent
• related concept
• content treatment
• next workflow

Use this to:

See the wider search landscape before narrowing the scope.

One seed topic can lead to many query branches, but only some of those branches belong in the site.

21. SERP Entity Harvest

Use this to extract repeated candidates from top ranking pages.

Short command:

text

Run SERP Entity Harvest on these SERP competitors.

Expanded prompt:

text

Run SERP Entity Harvest on these SERP competitors.

Use the source context first.

Extract repeated candidate entities, dominant terms, candidate attributes, repeated concepts, headings, table topics, FAQ topics, comparison angles, and visible entity signals from top ranking pages.

Return the output with these fields:
- candidate
- candidate type
- SERP occurrence count
- source URL count
- source placement
- related modifier
- candidate attribute
- consensus note
- source context fit
- keep, reject, or review
- next workflow route

Do not copy competitor structure.

Do not treat competitor coverage as required coverage by default.

Route the output into Competitive Coverage Snapshot, SERP Consensus Scan, Discovery Opportunity Matrix, or Information Gain.

Best for:

• competitor pages
• SERP exports
• content gap work
• topical map preparation
• brief intake

Output should include:

• candidate entity
• SERP occurrence
• source URL count
• related modifier
• consensus note
• next stage

Use this to:

Collect expected concepts, repeated subtopics, and common page signals from the SERP before deciding what to keep.

If the competitor scan shows repeated coverage, route the result into information gain work before briefing.

22. SERP Pattern Intake

Use this when the SERP shows repeated formats, page types, or query treatments.

Short command:

text

Run SERP Pattern Intake on this query.

Expanded prompt:

text

Run SERP Pattern Intake on this query.

Use the source context first.

Review the SERP for dominant page types, content formats, repeated section patterns, SERP features, answer formats, table use, FAQ use, comparison blocks, template blocks, local packs, product blocks, and missing angles.

Return the output with these fields:
- query
- dominant page type
- repeated format
- visible SERP feature
- repeated section
- repeated answer pattern
- missing angle
- source context fit
- content treatment note
- next workflow route

Do not write the content brief yet.

Do not finalize SERP feature strategy yet.

Route the output into Query Treatment Selection, Content Briefs, SERP Feature Planning, or Information Gain.

Best for:

• high value queries
• comparison topics
• commercial investigation
• PAA-heavy topics
• snippet targets

Output should include:

• dominant page type
• repeated format
• SERP feature
• repeated section
• missing angle
• treatment note

Use this to:

Choose the right downstream format.

A query might need a comparison page, guide, template, glossary page, use case page, or documentation page.

23. Competitor Entity Harvest

Use this to extract candidates from competitor pages without turning them into a map yet.

Short command:

text

Run Competitor Entity Harvest on these competitor URLs.

Expanded prompt:

text

Run Competitor Entity Harvest on these competitor URLs.

Use the source context first.

Extract raw candidate entities, candidate attributes, repeated concepts, page formats, proof points, comparison angles, product references, FAQ themes, and support topics from the competitor sources.

Return the output with these fields:
- extracted candidate
- candidate type
- source competitor
- source passage
- source placement
- repeated attribute
- coverage note
- confidence
- source context fit
- keep, reject, or review
- next workflow route

Do not copy competitor wording.

Do not copy competitor structure.

Use competitor pages as discovery evidence only.

Route the output into Competitive Coverage Snapshot, SERP Consensus Scan, Discovery Opportunity Matrix, or Information Gain.

Best for:

• top SERP pages
• competitor blogs
• comparison pages
• product pages
• docs pages

Output should include:

• extracted candidate
• source competitor
• source passage
• repeated attribute
• coverage note
• confidence

Use this to:

Find concepts your site may have missed without copying competitor structure.

Competitor data should guide discovery, not become the outline.

24. Competitive Coverage Snapshot

Use this when you need a short view of competitor coverage before deeper analysis.

Short command:

text

Run Competitive Coverage Snapshot on this SERP set.

Expanded prompt:

text

Run Competitive Coverage Snapshot on this SERP set.

Use the source context first.

Summarize what competitors commonly cover, what they overuse, what they miss, what formats they repeat, which proof points they rely on, and where the project may add useful differentiation.

Return the output with these fields:
- common coverage
- overused coverage
- missing coverage
- repeated format
- proof pattern
- differentiation seed
- source context fit
- risk note
- recommended next module
- next workflow route

Do not create the page outline yet.

Do not copy competitor sections.

Route the output into Information Gain, Content Briefs, SERP Pattern Intake, or Discovery Opportunity Matrix.

Best for:

• SERP gap prep
• information gain prep
• brief intake
• rewrite planning
• topic validation

Output should include:

• common coverage
• missing coverage
• overused coverage
• differentiation seed
• risk note
• next stage

Use this to:

See what everyone covers, what may be missing, and where the project can add useful differentiation.

This is a strong handoff into information gain work.

25. SERP Consensus Scan

Use this to identify what the SERP agrees on.

Short command:

text

Run SERP Consensus Scan on this query group.

Expanded prompt:

text

Run SERP Consensus Scan on this query group.

Use the source context first.

Identify repeated concepts, repeated claims, expected sections, repeated page formats, recurring examples, common definitions, common FAQ topics, common comparison angles, and expected answer patterns across the SERP.

Return the output with these fields:
- consensus concept
- repeated format
- expected section
- common claim
- common example type
- repeated FAQ topic
- gap
- differentiation note
- source context fit
- next workflow route

Separate expected coverage from repeated content.

Flag areas that need a stronger angle instead of another repeated answer.

Route the output into Information Gain, Content Briefs, SERP Feature Planning, or Discovery Opportunity Matrix.

Best for:

• brief preparation
• information gain intake
• competitor review
• SERP feature targeting
• page format selection

Output should include:

• consensus concept
• repeated format
• expected section
• common claim
• gap
• differentiation note

Use this to:

Separate expected coverage from repeated coverage.

Some SERP consensus helps show what users expect. Too much consensus can lead to duplicate content.

26. SERP Divergence Scan

Use this to find where top pages differ.

Short command:

text

Run SERP Divergence Scan on this query group.

Expanded prompt:

text

Run SERP Divergence Scan on this query group.

Use the source context first.

Find where top pages differ in intent, format, audience, page type, depth, angle, CTA path, proof, SERP feature targeting, and topic scope.

Return the output with these fields:
- divergence type
- source page
- different page type
- unique angle
- intent signal
- audience signal
- format signal
- risk note
- source context fit
- recommended next module
- next workflow route

Flag mixed SERPs.

Flag queries that should not become a page until the intent is clearer.

Route the output into Query Treatment Selection, SERP Pattern Intake, Content Briefs, or Information Gain.

Best for:

• uncertain search intent
• mixed SERPs
• commercial investigation
• comparison pages
• content strategy

Output should include:

• divergence type
• source page
• unique angle
• intent signal
• risk note
• next stage

Use this to:

Slow down when the SERP has unstable intent.

When top pages solve the query in different ways, MIRENA should not rush into page type, format, or brief structure.

27. Topical Authority Baseline

Use this to understand the depth needed before planning pages.

Short command:

text

Run Topical Authority Baseline on this topic.

Expanded prompt:

text

Run Topical Authority Baseline on this topic.

Use the source context first.

Review the topic, SERP set, current site coverage, competitor depth, related support areas, repeated entities, query branches, and missing support areas.

Return the output with these fields:
- topic
- current site coverage signal
- competitor depth signal
- expected support area
- repeated support area
- missing support area
- authority risk
- source context fit
- priority note
- next workflow route

Do not build the full topical map yet.

Use this as a baseline before page planning.

Route the output into Topical Maps and Planning, Content Briefs, or Discovery Opportunity Matrix.

Best for:

• new clusters
• hub planning
• competitor research
• site expansion
• authority gap reviews

Output should include:

• topic depth signal
• competitor depth signal
• expected support area
• missing support area
• priority note
• handoff note

Use this to:

Understand how much support a topic needs before pages are planned.

This helps MIRENA avoid thin maps and oversized maps.

28. Competitive Schema Scan

Use this when competitor pages rely on structured data or entity markup.

Short command:

text

Run Competitive Schema Scan on these competitor pages.

Expanded prompt:

text

Run Competitive Schema Scan on these competitor pages.

Use the source context first.

Review competitor pages for visible schema types, JSON-LD patterns, entity fields, FAQ markup patterns, HowTo markup patterns, Product or SoftwareApplication signals, Organization signals, breadcrumb patterns, sameAs cues, and repeated structured data fields.

Return the output with these fields:
- competitor source
- visible schema type
- entity field
- attribute pattern
- repeated structured data field
- schema gap note
- source context fit
- relevance to our page
- later schema cue
- next workflow route

Do not create final schema.

Do not add schema to an unapproved draft.

Use this as discovery input for later schema cues.

Route the output into Schema Cues after the draft is approved, or into Content Briefs if the schema pattern reveals required visible content.

Best for:

• product pages
• software pages
• FAQ-heavy pages
• how-to pages
• comparison pages

Output should include:

• schema type
• entity field
• attribute pattern
• competitor source
• gap note
• later schema cue

Use this to:

Collect competitive structured data signals that can inform later schema cues after the draft is approved.

This module does not create schema for your page.

29. Semantic Neighborhood Expansion

Use this when a topic needs adjacent concepts before mapping.

Short command:

text

Run Semantic Neighborhood Expansion on this seed topic.

Expanded prompt:

text

Run Semantic Neighborhood Expansion on this seed topic.

Use the source context first.

Expand the seed topic into nearby concepts, adjacent categories, related query paths, support concepts, comparison paths, process paths, feature paths, and concepts that should be rejected.

Return the output with these fields:
- adjacent concept
- relationship type
- related query path
- nearby category
- source context fit
- keep, reject, or review
- reject note
- risk note
- suggested next module
- next workflow route

Do not include adjacent concepts only because they are semantically near.

Reject concepts that drift outside the site’s scope.

Route the output into Query Network Expansion, Semantic Query Clustering, Topical Maps and Planning, or Discovery Opportunity Matrix.

Best for:

• early topical discovery
• weak seed topics
• semantic SEO planning
• query expansion
• cluster discovery

Output should include:

• adjacent concept
• relationship type
• query path
• category note
• reject note
• next workflow

Use this to:

Find adjacent ideas, then mark which ones fit and which ones should be rejected.

Use Semantic SEO when the discovery task needs stronger meaning, context, and topic fit before page planning.

30. Source Candidate Ranking

Use this when discovery pulls from many sources.

Short command:

text

Run Source Candidate Ranking on this discovery set.

Expanded prompt:

text

Run Source Candidate Ranking on this discovery set.

Use the source context first.

Rank each source by quality, relevance, freshness risk, evidence strength, extraction value, topic fit, intent fit, and risk of adding noise.

Return the output with these fields:
- source
- source type
- source quality score
- relevance score
- evidence strength
- extraction value
- freshness risk
- noise risk
- keep, hold, ignore, or review
- reason
- next workflow route

Do not treat every source equally.

Give more weight to sources that fit the project and less weight to sources that create drift.

Route the output into Candidate Weighting, SERP Entity Harvest, Competitor Entity Harvest, or Discovery Opportunity Matrix.

Best for:

• competitor URLs
• SERP exports
• file sets
• documents
• crawl extracts
• analytics files

Output should include:

• source
• source type
• quality score
• evidence value
• extraction value
• keep or ignore note

Use this to:

Protect the discovery layer from weak sources.

A weak source can pollute the entire workflow, so MIRENA should rank source quality before building the opportunity matrix.

31. Discovery Opportunity Matrix

Use this to turn raw discovery into decisions.

Short command:

text

Run Discovery Opportunity Matrix on this raw discovery output.

Expanded prompt:

text

Run Discovery Opportunity Matrix on this raw discovery output.

Use the source context first.

Turn raw findings into prioritized opportunities.

Review accepted candidates, rejected candidates, query paths, modifier groups, SERP patterns, competitor findings, source quality notes, and risk notes.

Return the output with these fields:
- opportunity
- opportunity type
- evidence source
- source context fit
- impact
- risk
- confidence
- effort
- recommended next workflow
- reason
- blocked items
- review-needed items

Group opportunities by next route:
- Topical Maps and Planning
- Content Briefs
- Drafting and Rewriting
- Entity SEO and Salience
- Internal Linking
- Information Gain
- SERP Feature Planning
- Schema Cues after approval
- Reject or hold

Do not create final briefs, maps, rewrites, or schema.

Route the output into Discovery Handoff.

Best for:

• final raw discovery output
• handoff to topical mapping
• handoff to content briefs
• handoff to entity SEO
• content strategy

Output should include:

• opportunity
• opportunity type
• evidence source
• impact
• risk
• recommended next stage

Use this to:

Turn raw findings into next-step choices for mapping, briefing, rewriting, internal links, information gain, or rejection.

Discovery is only useful when it becomes action.

32. Entity Universe Package

Use this as the final discovery package before downstream workflows.

Short command:

text

Run Entity Universe Package from this discovery output.

Expanded prompt:

text

Run Entity Universe Package from this discovery output.

Use the source context first.

Package the accepted raw candidates, rejected candidates, review-needed candidates, query paths, modifier groups, candidate attributes, source notes, SERP signals, competitor signals, and handoff notes.

Return the output with these fields:
- accepted candidate
- candidate type
- source evidence
- supporting query path
- related modifier group
- candidate attribute
- SERP signal
- competitor signal
- rejection list
- review-needed list
- source context fit
- handoff note
- next workflow route

Do not create the finished entity map.

Do not run salience.

Do not create schema.

Route the output into Topical Maps and Planning, Content Briefs, Entity SEO and Salience, Drafting and Rewriting, or Information Gain.

Best for:

• raw discovery completion
• topical map intake
• brief intake
• entity SEO intake
• rewrite intake

Output should include:

• accepted candidates
• rejected candidates
• query paths
• modifier groups
• SERP signals
• source notes
• handoff notes

Use this to:

Create the organized raw discovery handoff.

When the package needs entity structure, route it into Entity SEO or the Entity SEO and Salience prompt guide.

33. Page Archetype Seed Discovery

Use this when raw discovery should suggest page types but not build the map yet.

Short command:

text

Run Page Archetype Seeds from this discovery set.

Expanded prompt:

text

Run Page Archetype Seeds from this discovery set.

Use the source context first.

Look for signals that suggest likely page archetypes, user jobs, page roles, and downstream routes.

Review query intent, modifiers, SERP patterns, competitor formats, existing site pages, user stages, content formats, and commercial goals.

Return the output with these fields:
- likely page archetype
- user job
- intent signal
- source evidence
- possible page role
- source context fit
- overlap risk
- keep, reject, or review
- recommended downstream workflow
- reason

Do not build the processed topical map yet.

Use this only to seed page type decisions.

Route the output into Topical Maps and Planning, Content Briefs, Docs planning, Use Case planning, Comparison planning, or Templates and Examples planning.

Best for:

• topical map prep
• use case planning
• comparison planning
• docs planning
• template planning

Output should include:

• page archetype
• user job
• intent signal
• evidence source
• route to next workflow
• risk note

Use this to:

Identify likely docs pages, comparison pages, use case pages, templates, examples, hubs, or support pages before a processed map exists.

This is a page type signal, not the final map.

34. Discovery Handoff

Use this to move raw discovery output into the next workflow.

Short command:

text

Run Discovery Handoff for this raw discovery output.

Expanded prompt:

text

Run Discovery Handoff for this raw discovery output.

Use the source context first.

Route each useful finding into the correct next workflow.

Review accepted candidates, rejected candidates, review-needed candidates, query clusters, modifier groups, SERP findings, competitor findings, source quality notes, opportunity matrix items, and blocked items.

Return the output with these fields:
- finding
- finding type
- evidence source
- source context fit
- route to Topical Maps and Planning
- route to Content Briefs
- route to Drafting and Rewriting
- route to Entity SEO and Salience
- route to Internal Linking
- route to Information Gain
- route to SERP Feature Planning
- route to Schema Cues after approval
- blocked item
- reason
- owner or next action
- handoff note

Do not leave raw findings without a route.

Do not send off-scope candidates downstream.

Flag anything that needs review before the next workflow starts.

Return a clean handoff package.

Best for:

• end of raw discovery
• large discovery outputs
• multi-team handoff
• audit work
• workflow routing

Output should include:

• route to topical mapping
• route to content brief
• route to rewrite
• route to entity SEO
• route to internal linking
• route to information gain
• blocked items
• review notes

Use this to:

Turn raw research into a routed workflow package.

Discovery Handoff tells MIRENA where every useful finding should go next.

Which Raw Discovery Module Should You Run First?

Start with Source Context Check if the project boundary is not clear.

Start with Discovery Asset Review if you have many files.

Start with Corpus Scan if you have a content export or page set.

Start with NER Pass if you have text and need candidate entities.

Start with Query Intent Classification if you have a keyword export.

Start with SERP Entity Harvest if you have competitor pages.

Start with Discovery Opportunity Matrix if a discovery pass already exists and you need decisions.

Start with Discovery Handoff when you need to route raw findings into the next MIRENA workflow.

Common Raw Discovery Starting Points

I Have a Keyword Export

Start with source context, then classify the queries.

Prompt:

text

Run Query Intent Classification on this keyword export.

Use the source context first.

Classify each query by primary intent, secondary intent, user stage, likely page type, and content treatment.

Return the output with these fields:
- query
- primary intent
- secondary intent
- user stage
- page type recommendation
- content treatment
- SERP format note
- source context fit
- keep, merge, section, FAQ, comparison, anchor, or reject
- reason
- next workflow route

Route the final output into Topical Maps and Planning if the query group needs structure.

Then run:

text

Run Semantic Query Clustering on this query set.

Use the source context first.

Group queries by meaning, intent layer, user job, page type, modifier pattern, and shared concept.

Return the output with these fields:
- cluster name
- included queries
- shared concept
- dominant intent
- secondary intent
- likely page type
- page vs section risk
- overlap risk
- source context fit
- next workflow route

Route the output into Query Treatment Selection or Topical Maps and Planning.

Then run:

text

Run Discovery Opportunity Matrix on this raw discovery output.

Use the source context first.

Turn raw findings into prioritized opportunities.

Return opportunities grouped by:
- Topical Maps and Planning
- Content Briefs
- Drafting and Rewriting
- Entity SEO and Salience
- Internal Linking
- Information Gain
- SERP Feature Planning
- Schema Cues after approval
- Reject or hold

Include impact, risk, confidence, effort, and reason for each route.

Route the result into Topical Maps + Planning.

I Have a Sitemap or URL List

Start with source context, then scan the corpus.

Prompt:

text

Run Corpus Scan on this page set.

Use the source context first.

Scan the content before deeper extraction.

Return recurring terms, recurring concepts, repeated phrases, weak signals, candidate entities, modifier candidates, source locations, confidence notes, and suggested next modules.

Do not turn the output into a topical map.

Route the output into NER Pass, Concept Harvest, Frequency Signal Scan, or Placement Signal Scan.

Then run:

text

Run NER Pass on this asset.

Use the source context first.

Extract named entity candidates from the asset.

Return candidate entities, entity types, source locations, frequency signals, placement signals, confidence levels, source context fit, keep or reject notes, and next workflow routes.

Do not build an entity map.

Route the output into Entity Type Classification, Candidate Weighting, or Entity Universe Package.

Then run:

text

Run Entity Universe Package from this discovery output.

Use the source context first.

Package accepted candidates, rejected candidates, review-needed candidates, query paths, modifier groups, candidate attributes, SERP signals, source notes, and handoff notes.

Route the output into Topical Maps and Planning, Content Briefs, Entity SEO and Salience, Drafting and Rewriting, or Information Gain.

Route the result into Topical Maps + Planning or Entity SEO.

I Have Competitor URLs

Start with source context, then harvest competitor signals.

Prompt:

text

Run Competitor Entity Harvest on these competitor URLs.

Use the source context first.

Extract raw candidate entities, candidate attributes, repeated concepts, page formats, proof points, comparison angles, product references, FAQ themes, and support topics from the competitor sources.

Return extracted candidates, source competitor, source passage, source placement, repeated attributes, coverage notes, confidence, source context fit, and next workflow route.

Do not copy competitor wording or structure.

Then run:

text

Run Competitive Coverage Snapshot on this SERP set.

Use the source context first.

Summarize common coverage, overused coverage, missing coverage, repeated formats, proof patterns, differentiation seeds, risk notes, and recommended next modules.

Route the output into Information Gain, Content Briefs, SERP Pattern Intake, or Discovery Opportunity Matrix.

Then run:

text

Run SERP Consensus Scan on this query group.

Use the source context first.

Identify repeated concepts, repeated claims, expected sections, repeated page formats, recurring examples, common definitions, common FAQ topics, common comparison angles, and expected answer patterns.

Separate expected coverage from repeated content.

Route the output into Information Gain, Content Briefs, SERP Feature Planning, or Discovery Opportunity Matrix.

Route the result into information gain work, Topical Maps + Planning, or Content Briefs.

I Have GSC Query Data

Start with source context, then classify and expand.

Prompt:

text

Run Query Modifier Scan on this query set.

Use the source context first.

Scan query wording for modifiers that change page type, intent, format, audience, product stage, local need, comparison need, trust need, or next step.

Return modifier classes, example queries, repeated patterns, intent shifts, page type signals, source context fit, keep or reject notes, and next workflow routes.

Then run:

text

Run Query Treatment Selection on this query list.

Use the source context first.

Decide how each query should be handled before any new page is created.

Choose from dedicated page, section, FAQ, list, comparison block, table, anchor text target, internal link target, template, example, merge into existing page, or reject.

Return the reason and next workflow route for every query.

Then run:

text

Run Query Network Expansion on this seed topic.

Use the source context first.

Expand the seed topic into related query paths, intent branches, user stages, support questions, comparison paths, feature paths, process paths, and adjacent topics.

Reject branches that drift outside the source context.

Route the output into Semantic Query Clustering, Query Treatment Selection, Topical Maps and Planning, or Content Briefs.

Route the result into Topical Maps + Planning, Content Briefs, or internal links.

I Have a Draft Folder

Start with source context, then scan and extract.

Prompt:

text

Run Corpus Scan on this draft set.

Use the source context first.

Scan the drafts for recurring terms, recurring concepts, repeated phrases, weak signals, candidate entities, modifier candidates, low-confidence terms, overused phrases, and extraction notes.

Return source locations and recommended next modules.

Do not rewrite the drafts yet.

Then run:

text

Run Placement Signal Scan on this asset.

Use the source context first.

Check where raw candidates appear across titles, headings, openings, body sections, tables, FAQs, anchors, CTA sections, and schema notes.

Return placement locations, placement strength, repeated patterns, source context fit, confidence, and next workflow route.

Use placement only as a discovery signal.

Then run:

text

Run Discovery Handoff for this raw discovery output.

Use the source context first.

Route findings into Topical Maps and Planning, Content Briefs, Drafting and Rewriting, Entity SEO and Salience, Internal Linking, Information Gain, SERP Feature Planning, Schema Cues after approval, or Reject.

Flag anything that needs review before the next workflow starts.

Route the result into Drafting + Rewriting if the next job is page repair.

I Have a Broad Seed Topic

Start with source context, then expand the topic carefully.

Prompt:

text

Run Semantic Neighborhood Expansion on this seed topic.

Use the source context first.

Expand the seed topic into nearby concepts, adjacent categories, related query paths, support concepts, comparison paths, process paths, feature paths, and concepts that should be rejected.

Reject concepts that drift outside the site’s scope.

Return adjacent concepts, relationship types, query paths, category notes, reject notes, risks, and next workflow routes.

Then run:

text

Run Latent Intent Discovery on this topic.

Use the source context first.

Find hidden user needs implied by the topic, query set, SERP, competitors, modifiers, product context, or user stage.

Return latent intents, source signals, related query patterns, user needs, likely page types, content treatments, confidence notes, and next workflow routes.

Do not create new pages from latent intent alone.

Then run:

text

Run Page Archetype Seeds from this discovery set.

Use the source context first.

Look for signals that suggest likely page archetypes, user jobs, page roles, and downstream routes.

Return likely page archetypes, user jobs, intent signals, source evidence, possible page roles, overlap risks, keep or reject notes, and recommended downstream workflows.

Route the result into Topical Maps + Planning.

Output Review Checklist for Raw Discovery

Before moving downstream, check the raw discovery output for:

• source context fit
• accepted candidates
• rejected candidates
• review-needed candidates
• useful concept groups
• useful modifier groups
• candidate attributes
• query intent labels
• query treatment suggestions
• latent query paths
• query clusters
• SERP entity patterns
• competitor coverage notes
• consensus signals
• divergence signals
• source quality notes
• opportunity matrix
• handoff route
• blocked topics
• uncertain candidates

Do not move into topical mapping, content briefs, entity SEO, rewriting, internal links, or schema if the raw discovery package still contains unclear, off-scope, or unsupported candidates.

How Raw Discovery Connects to Other MIRENA Workflows

Raw Discovery is the intake and discovery layer.

It feeds other workflows but does not replace them.

A discovery package can feed Topical Maps + Planning when the next job is structure.

A query classification output can feed Content Briefs when the next job is writer instruction.

A competitor coverage snapshot can feed information gain work when the next job is useful differentiation.

A corpus scan can feed Drafting + Rewriting when the next job is repairing existing pages.

A semantic discovery package can feed Semantic SEO when the next job is meaning, coverage, and topic fit.

An entity universe package can feed Entity SEO when the next job is entity structure.

Use MIRENA workflow when you need to decide the next route. Use MIRENA outputs when you need the final discovery package to follow a set output structure.

Raw Discovery Mistakes to Avoid

Mistake 1: Running Discovery Without Source Context

Raw discovery can collect too much.

If source context is missing, MIRENA may pull in candidates that do not belong on the site.

Start with source context every time.

Mistake 2: Treating Raw Discovery as a Finished Map

Raw discovery is not a processed topical map.

It finds signals.

A later workflow decides page roles, hierarchy, publishing order, page vs section decisions, and internal link routes.

Mistake 3: Keeping Every Candidate

A candidate is not a recommendation.

Use Candidate Weighting, Source Candidate Ranking, and Discovery Opportunity Matrix before moving findings downstream.

Mistake 4: Copying Competitor Coverage

Competitor harvests are for evidence, not imitation.

Use competitor findings to understand consensus, gaps, and opportunities. Then route the output into information gain or content briefing.

Mistake 5: Skipping Query Treatment

A query may need a page, section, FAQ, anchor, comparison, list, or rejection.

Run Query Treatment Selection before creating new pages from keyword data.

Mistake 6: Sending Unclear Findings Downstream

Do not hand off a discovery package full of uncertain candidates.

Mark accepted, rejected, and review-needed items before moving into topical mapping or briefing.

Mistake 7: Repeating Entity SEO Work Too Early

Raw discovery collects raw candidates.

Entity SEO and Salience organizes, scores, places, repairs, and prepares entity structure.

Do not turn this workflow into an entity salience workflow.

Example Raw Discovery Workflow for a Keyword Export

Use this when you have raw keyword data.

text

Run Query Intent Classification on this keyword export.

Use the source context first.

Classify each query by primary intent, secondary intent, user stage, likely page type, and content treatment.

Return the output with these fields:
- query
- primary intent
- secondary intent
- user stage
- page type recommendation
- content treatment
- SERP format note
- source context fit
- keep, merge, section, FAQ, comparison, anchor, or reject
- reason
- next workflow route

Route the final output into Topical Maps and Planning if the query group needs structure.

Then:

text

Run Query Modifier Scan on this query set.

Use the source context first.

Scan query wording for modifiers that change page type, intent, format, audience, product stage, local need, comparison need, trust need, or next step.

Return modifier classes, example queries, repeated patterns, intent shifts, page type signals, source context fit, keep or reject notes, and next workflow routes.

Then:

text

Run Semantic Query Clustering on this query set.

Use the source context first.

Group queries by meaning, intent layer, user job, page type, modifier pattern, and shared concept.

Return cluster names, included queries, shared concepts, dominant intent, secondary intent, likely page type, page vs section risk, overlap risk, source context fit, and next workflow route.

Then:

text

Run Discovery Handoff for this raw discovery output.

Use the source context first.

Route each useful finding into the correct next workflow.

Return findings grouped by Topical Maps and Planning, Content Briefs, Drafting and Rewriting, Entity SEO and Salience, Internal Linking, Information Gain, SERP Feature Planning, Schema Cues after approval, and Reject.

Flag review-needed items before handoff.

Example Raw Discovery Workflow for Competitor Research

Use this when you have competitor URLs or SERP exports.

text

Run SERP Entity Harvest on these SERP competitors.

Use the source context first.

Extract repeated candidate entities, dominant terms, candidate attributes, repeated concepts, headings, table topics, FAQ topics, comparison angles, and visible entity signals from top ranking pages.

Return candidate type, SERP occurrence count, source URL count, source placement, related modifiers, candidate attributes, consensus notes, source context fit, keep or reject notes, and next workflow route.

Do not copy competitor structure.

Then:

text

Run Competitive Coverage Snapshot on this SERP set.

Use the source context first.

Summarize common coverage, overused coverage, missing coverage, repeated formats, proof patterns, differentiation seeds, risk notes, and recommended next modules.

Route the output into Information Gain, Content Briefs, SERP Pattern Intake, or Discovery Opportunity Matrix.

Then:

text

Run SERP Divergence Scan on this query group.

Use the source context first.

Find where top pages differ in intent, format, audience, page type, depth, angle, CTA path, proof, SERP feature targeting, and topic scope.

Return divergence type, source page, unique angle, intent signal, audience signal, format signal, risk note, source context fit, and next workflow route.

Then:

text

Run Discovery Opportunity Matrix on this raw discovery output.

Use the source context first.

Turn raw findings into prioritized opportunities.

Return opportunity type, evidence source, source context fit, impact, risk, confidence, effort, recommended next workflow, reason, blocked items, and review-needed items.

Example Raw Discovery Workflow for an Existing Site

Use this when you have a sitemap, crawl, or page inventory.

text

Run Discovery Asset Review on these files.

Use the source context first.

Review each uploaded file, export, URL list, report, brief, sitemap, crawl, search file, analytics file, behavior note, competitor source, or previous MIRENA output.

Return asset type, source quality, freshness risk, discovery value, best use, noise risk, missing fields, recommended processing order, keep or ignore notes, and next workflow route.

Then:

text

Run Corpus Scan on this content set.

Use the source context first.

Scan the content before running deeper extraction.

Return recurring terms, recurring concepts, repeated phrases, page themes, weak signals, overused language, underdeveloped concepts, candidate entities, modifier candidates, and content sections that need deeper review.

Then:

text

Run Placement Signal Scan on this asset.

Use the source context first.

Check where raw candidates appear across titles, meta titles, headings, openings, body sections, tables, FAQs, captions, internal anchors, schema notes, and CTA sections.

Return candidate, source page, placement location, placement strength, repeated placement pattern, source context fit, importance note, extraction confidence, suggested next module, and next workflow route.

Then:

text

Run Entity Universe Package from this discovery output.

Use the source context first.

Package accepted candidates, rejected candidates, review-needed candidates, query paths, modifier groups, candidate attributes, source notes, SERP signals, competitor signals, and handoff notes.

Route the output into Topical Maps and Planning, Content Briefs, Entity SEO and Salience, Drafting and Rewriting, or Information Gain.

Example Raw Discovery Workflow for Search Console Data

Use this when you have GSC pages and query exports.

text

Run Query Modifier Scan on this query set.

Use the source context first.

Scan query wording for modifiers that change page type, intent, format, audience, product stage, local need, comparison need, trust need, or next step.

Return modifier classes, example queries, repeated patterns, intent shifts, page type signals, source context fit, keep or reject notes, and next workflow routes.

Then:

text

Run Query Treatment Selection on this query list.

Use the source context first.

Decide how each query should be handled before any new page is created.

Choose from dedicated page, section, FAQ, list, comparison block, table, anchor text target, internal link target, template, example, merge into existing page, or reject.

Return the reason and next workflow route for every query.

Then:

text

Run Query Network Expansion on this seed topic.

Use the source context first.

Expand the seed topic into related query paths, intent branches, user stages, support questions, comparison paths, feature paths, process paths, and adjacent topics.

Reject branches that drift outside the source context.

Route the output into Semantic Query Clustering, Query Treatment Selection, Topical Maps and Planning, or Content Briefs.

Then route the output into topical mapping, briefs, or internal links.

Example Raw Discovery Workflow for SERP Feature Research

Use this when the SERP format is unclear.

text

Run SERP Pattern Intake on this query.

Use the source context first.

Review the SERP for dominant page types, content formats, repeated section patterns, SERP features, answer formats, table use, FAQ use, comparison blocks, template blocks, local packs, product blocks, and missing angles.

Return dominant page type, repeated format, visible SERP feature, repeated section, repeated answer pattern, missing angle, source context fit, content treatment note, and next workflow route.

Then:

text

Run SERP Consensus Scan on this query group.

Use the source context first.

Identify repeated concepts, repeated claims, expected sections, repeated page formats, recurring examples, common definitions, common FAQ topics, common comparison angles, and expected answer patterns.

Separate expected coverage from repeated content.

Route the output into Information Gain, Content Briefs, SERP Feature Planning, or Discovery Opportunity Matrix.

Then:

text

Run Discovery Handoff for this raw discovery output.

Use the source context first.

Route each useful finding into the correct next workflow.

Return findings grouped by Topical Maps and Planning, Content Briefs, Drafting and Rewriting, Entity SEO and Salience, Internal Linking, Information Gain, SERP Feature Planning, Schema Cues after approval, and Reject.

Flag anything that needs review before the next workflow starts.

FAQs About Raw Semantic Discovery Prompts in MIRENA

What is Raw Semantic Discovery in MIRENA?

Raw Semantic Discovery is the upstream workflow that collects semantic evidence before MIRENA creates maps, briefs, rewrites, internal links, information gain passes, SERP feature notes, or schema notes.

It extracts candidates, concepts, modifiers, query signals, SERP patterns, competitor coverage, and opportunity notes.

What should I run first?

Start with Source Context Check.

If the source context is already strong, run Discovery Asset Review when you have many files, Corpus Scan when you have a page set, Query Intent Classification when you have keywords, or SERP Entity Harvest when you have competitor pages.

Is Raw Semantic Discovery the same as Entity SEO?

No.

Raw Semantic Discovery collects candidates.

Entity SEO and Salience workflows organize, score, place, repair, and prepare entity structure.

Is Raw Semantic Discovery the same as topical mapping?

No.

Raw discovery gathers signals. Topical mapping turns selected signals into a governed page structure.

Can I use keyword exports with this workflow?

Yes.

Use Query Intent Classification, Query Modifier Scan, Semantic Query Clustering, and Query Treatment Selection to prepare keyword data before topical mapping.

Can I use competitor pages with this workflow?

Yes.

Use Competitor Entity Harvest, SERP Entity Harvest, Competitive Coverage Snapshot, SERP Consensus Scan, and SERP Divergence Scan.

Competitor data should guide discovery, not copy structure.

Can I use GSC data with this workflow?

Yes.

Use GSC query exports for Query Modifier Scan, Query Intent Classification, Query Treatment Selection, and Query Network Expansion.

Use GSC page data to find which existing pages may need mapping, briefing, rewriting, or internal link support.

Can I use GA4 or behavior data with this workflow?

Yes.

Raw Semantic Discovery can use analytics and behavior notes as evidence. Those files help MIRENA understand which pages attract users, where users stall, and which paths may need repair.

What does the Entity Universe Package do?

The Entity Universe Package collects accepted candidates, rejected candidates, query paths, modifier groups, SERP signals, source notes, and handoff notes.

It prepares raw evidence for downstream workflows.

What should I do after Raw Semantic Discovery?

Move the output into the correct next workflow.

Use Topical Maps + Planning when the next job is structure.

Use Content Briefs when the next job is writer instruction.

Use Entity SEO when the next job is entity structure.

Use Drafting + Rewriting when the next job is page repair.

Use internal linking when the next job is route building.

Use information gain work when the next job is useful differentiation.

Copy to Master Context

text

PAGE CREATED:
https://semantecseo.com/docs/raw-semantic-discovery-prompts/

PARENT PAGE:
https://semantecseo.com/docs/getting-started/

READER-FACING WORKFLOW FAMILY:
Raw Semantic Discovery

INTERNAL WORKFLOW LABEL:
P3 — backend only

PRIMARY ENTITY:
Raw semantic discovery prompts for MIRENA

PAGE ROLE:
Prompt playbook for collecting raw semantic evidence before topical mapping, content briefs, entity SEO, rewriting, internal links, information gain, SERP feature planning, or schema notes.

CORE RULES:
Source context comes first.
Use short action-led prompts.
Use expanded prompts for large, messy, risky, or handoff-ready tasks.
Do not require full module titles.
Do not repeat Entity SEO and Salience prompts.
Raw discovery collects signals.
Later workflows decide structure, salience, briefs, rewrites, links, and schema.
Do not move downstream with unclear, off-scope, or unsupported candidates.
Avoid blacklisted terms and variants.
Use contextual internal links.

MODULE COUNT:
34

KEY MODULES:
Source Context Check
Discovery Asset Review
Corpus Scan
NER Pass
Concept Harvest
Entity Type Classification
Candidate Weighting
Frequency Signal Scan
Placement Signal Scan
Modifier Harvest
Attribute Candidate Harvest
Facet Intent Extraction
Query Intent Classification
Query Modifier Scan
Query Treatment Selection
Synthetic Query Generation
Synthetic Query Classification
Semantic Query Clustering
Latent Intent Discovery
Query Network Expansion
SERP Entity Harvest
SERP Pattern Intake
Competitor Entity Harvest
Competitive Coverage Snapshot
SERP Consensus Scan
SERP Divergence Scan
Topical Authority Baseline
Competitive Schema Scan
Semantic Neighborhood Expansion
Source Candidate Ranking
Discovery Opportunity Matrix
Entity Universe Package
Page Archetype Seed Discovery
Discovery Handoff

PRIMARY INTERNAL LINKS:
Getting Started with MIRENA
Source context template
Source Context
MIRENA inputs
MIRENA workflow
MIRENA outputs
Topical Maps + Planning
Content Briefs
Drafting + Rewriting
Information Gain
Semantic SEO
Entity SEO

NEXT WORKFLOWS:
Topical Maps + Planning
Content Briefs
Drafting and Rewriting
Entity SEO and Salience
Internal Linking
Information Gain
SERP Feature Planning
Schema Cues After Approval