From create-provider
Automatically research and create LeanIX Provider fact sheets with verified data. Use this skill whenever the user mentions creating providers, adding providers, researching providers, or says phrases like "Create Provider for X", "Add Provider X", "Research provider X", or provides a provider name/URL to add to the catalog. Also trigger when user mentions provider data collection, provider research, or vendor catalog management. This skill automates the entire workflow - always use it rather than doing ad-hoc research.
How this skill is triggered — by the user, by Claude, or both
Slash command
/create-provider:create-providerThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill automates the complete Provider fact sheet creation workflow using parallel research, cross-verification, and quality checks.
AGENT_INSTRUCTIONS.mdFIELD_REFERENCE.mdIMPLEMENTATION_SUMMARY.mdMARKETING_LANGUAGE_FILTER.mdPROVIDER_CREATION_GUIDE.mdQUICK_START.mdREADME.mdSKILL_INVOCATION_PATTERNS.mdWORKFLOW_V2.mdfield_config.jsonguidelines/Provider_Alias_Discovery_Guidelines.mdguidelines/Provider_Classification_Definitions.mdguidelines/Provider_Description_Guidelines.mdguidelines/Provider_Headquarters_Address_Guidelines.mdguidelines/Provider_URL_Validation_Guidelines.mdlib/__init__.pylib/parallel_researcher.pylib/perplexity_client.pylib/provider_researcher.pylib/verification_agent.pyThis skill automates the complete Provider fact sheet creation workflow using parallel research, cross-verification, and quality checks.
This is a RIGID skill. Follow the workflow exactly. No ad-hoc queries. No improvisation.
Before doing ANYTHING, complete this checklist:
WORKFLOW_V2.md in this directory (complete workflow)guidelines/ directory:
Provider_Classification_Definitions.mdProvider_URL_Validation_Guidelines.mdProvider_Headquarters_Address_Guidelines.mdProvider_Alias_Discovery_Guidelines.mdProvider_Description_Guidelines.mdmcp__perplexity__perplexity_search)mcp__LeanIX_MCP_Server_Remote__)lib/provider_researcher.py, lib/parallel_researcher.py, lib/verification_agent.py (these show the correct query formats to use)If you're about to skip reading WORKFLOW_V2.md or make ad-hoc Perplexity calls, STOP.
Ask yourself: "Why am I not following the documented workflow? What step did I skip?"
Then: Go back, read the documentation, start from Step 1.
Trigger when user says:
Examples:
Automates Provider fact sheet creation through a 4-step workflow:
8 Fields Auto-Populated:
IMPORTANT: Read WORKFLOW_V2.md for the complete parallel research protocol.
The Python modules in lib/ are query templates that show you exactly what to ask. Read them first to understand the correct query structure, then make parallel API calls.
Step 1a: Read the Query Templates
Before making ANY queries, read these files to see the exact query format:
lib/provider_researcher.py - Shows the 5 Perplexity query methodslib/parallel_researcher.py - Shows WebFetch extraction promptsStep 1b: Execute Parallel Research (Use the Templates)
Now make 8 simultaneous tool calls using the query formats from the modules:
Perplexity Queries (5 calls):
get_url_research_query() format from provider_researcher.pyget_category_research_query() formatget_aliases_research_query() formatget_headquarters_research_query() formatget_description_research_query() formatWebFetch Queries (3 calls):
6. Homepage - Use _get_homepage_extraction_prompt() format from parallel_researcher.py
7. About Page - Use _get_about_extraction_prompt() format
8. Contact Page - Use _get_contact_extraction_prompt() format
Example Pattern:
# Read the template first
Read lib/provider_researcher.py
# Then make the call using that query format
mcp__perplexity__perplexity_search(
query="<exact query from get_url_research_query() method>"
)
DO NOT improvise queries. DO NOT skip reading the modules. The templates ensure consistency with the guidelines.
IMPORTANT: Read WORKFLOW_V2.md for complete verification protocol.
Use lib/verification_agent.py to:
Marketing Language Filter: For ALL descriptions, check for buzzwords:
MARKETING_LANGUAGE_FILTER.md for complete listOutput: Verified data with confidence scores per field
Before creating fact sheet, verify:
If any check fails, present options:
# 4.1 Create fact sheet via LeanIX MCP
fact_sheet_id = mcp__LeanIX_MCP_Server_Remote__create_fact_sheet(
name=provider_name,
type="Provider"
)
# 4.2 Update custom fields via Python CLI
# Run: python create-provider/main.py update --fact-sheet-id {id} --type Provider --fields '{...}'
update_fields = {
"homePageUrl": verified_data.url,
"providerCategory": verified_data.category,
"aliases": verified_data.aliases,
"headquartersAddress": verified_data.headquarters,
"description": verified_data.description,
"collectionStatus": "inReview",
"asOfDate": today's date,
"deprecated": "no"
}
# 4.3 Return URL
return f"https://{subdomain}.leanix.net/{workspace}/factsheet/Provider/{fact_sheet_id}"
Before researching ANY field, read the corresponding guideline:
File: guidelines/Provider_Classification_Definitions.md
3 categories with decision tree:
Decision order matters: Check Individual first, then Enterprise, then Community.
File: guidelines/Provider_URL_Validation_Guidelines.md
NEVER HALLUCINATE URLS. Core principles:
File: guidelines/Provider_Headquarters_Address_Guidelines.md
Priority order:
Partial data is OK: "San Francisco, CA, United States" is better than guessing full street address.
File: guidelines/Provider_Alias_Discovery_Guidelines.md
7 alias types to check:
File: guidelines/Provider_Description_Guidelines.md
Requirements:
Executes single-provider research via Perplexity with 5 specialized queries.
Use when: You need to research one provider using Perplexity MCP.
Coordinates parallel research across Perplexity + WebFetch.
Use when: Starting Step 1 of the workflow.
Compares results from multiple sources and resolves conflicts.
Use when: Starting Step 2 of the workflow (after parallel research completes).
Updates custom fields via GraphQL API.
Use when: Step 4, after fact sheet is created via LeanIX MCP.
Command format:
cd create-provider
python main.py update \
--fact-sheet-id "uuid-here" \
--type Provider \
--fields '{"homePageUrl": "...", "providerCategory": "...", ...}'
After completing workflow, present results like this:
✓ Provider created successfully!
Provider: TeamSmart AI
Overall Confidence: 88% (HIGH)
Verified Data:
├─ Homepage URL: https://teamsmart.ai/ ✓✓ (both sources)
├─ Category: Enterprise ✓✓ (both sources)
├─ Aliases: TeamSmart.ai ✓ (perplexity verified)
├─ Headquarters: San Francisco, CA, US ✓✓ (both sources)
└─ Description: 44 words ✓✓ (official website, marketing language removed)
Conflicts Resolved: 2
- Aliases: Used Perplexity (WebFetch found none)
- Description: Rewrote to remove "seamless", "transforms", "enhance"
URL: https://demo-eu-10.leanix.net/ltlsCollectionTesting/factsheet/Provider/{id}
✓ WebFetch: Successfully scraped website
✗ Perplexity: MCP connection failed
Action: Use WebFetch data only
Confidence: Downgraded to MEDIUM (single source)
Proceed: YES (sufficient data from official website)
✗ WebFetch: Website blocking/timeout
✓ Perplexity: Got results with citations
Action: Use Perplexity data only
Confidence: MEDIUM to HIGH (depends on citation quality)
Proceed: YES (Perplexity provides authoritative sources)
✗ Perplexity: MCP error
✗ WebFetch: Website not accessible
Action: Cannot proceed
User prompt: "Unable to research [Provider]. Please provide data manually or try again later."
⚠ Perplexity: Category = "Individual"
⚠ WebFetch: Category = "Enterprise"
Action: Run verification query to Perplexity
If still unclear: Present evidence to user and ask them to choose
❌ Making ad-hoc Perplexity queries instead of using provider_researcher.py ❌ Skipping WORKFLOW_V2.md and improvising the workflow ❌ Not reading guidelines before researching fields ❌ Hallucinating URLs instead of finding authoritative sources ❌ Using marketing language in descriptions (always filter buzzwords) ❌ Creating fact sheet before quality check (validate confidence first) ❌ Bypassing verification step (always cross-check sources)
Without this workflow, you get:
With this workflow, you get:
Speed: 5-10 seconds total (parallelized) Accuracy: 88%+ confidence typical Reliability: Fallback if one source fails
Before creating a Provider, ask yourself:
If answer to ANY question is "no" → STOP and read documentation first.
WORKFLOW_V2.md - Complete workflow with examplesguidelines/ - 5 field-specific guidelinesMARKETING_LANGUAGE_FILTER.md - Buzzword detectionFIELD_REFERENCE.md - LeanIX field specificationsREADME.md - Python CLI usagelib/ - Python modules for research and verificationWhen user says "Create Provider for [name]":
Remember: This is a RIGID workflow. Follow it exactly. No shortcuts.
npx claudepluginhub vineetgoyal1/leanix-catalog-research-marketplace --plugin create-providerQueries the ServiceGraph API to find, shortlist, and enrich US management consulting firms by industry, service, location, size, and ratings.
Researches target companies for sales prospecting using Browserbase Search API and a Plan→Research→Synthesize pattern with depth modes. Outputs scored reports and CSV.
Finds company and contact data from 150M+ companies and 800M+ professionals for lead generation, prospect research, and talent identification.