Enterprise search

Skill cxcscmu/SkillLearnBench/skills/b4-skill-creator-claude-haiku-4-5/enterprise-information-search/enterprise-search

Search and query enterprise JSON data to find specific information like employee IDs, document references, URLs, and insights. Use this skill when you need to locate answers to business questions such as finding authors of reports, team members with specific insights, or shared URLs. Handles multi-step queries, filtering, and deduplication of results.From its SKILL.md

Install

npx -y skills add cxcscmu/SkillLearnBench --skill enterprise-search

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

2.4 KB, 550 tokens by cl100k_base, as published. Nobody here has run it

Enterprise Data Search Strategies

Finding Document Authors and Reviewers

Pattern: Search for document name + extract userId from messages mentioning it

# Find all messages mentioning a specific report
grep -i "market research report\|market research" /root/DATA/products/ContentForce.json

# Extract from those messages: the userId (author/reviewer)
grep -i "market research" /root/DATA/products/ContentForce.json | grep -o '"userId":"[^"]*"' | cut -d'"' -f4

Finding Team Members with Specific Insights

Pattern: Search for keywords related to competitor products or strengths/weaknesses

# Search for competitor mentions
grep -i "competitor\|strengths\|weaknesses\|advantages\|disadvantages" /root/DATA/products/ContentForce.json

# Extract userIds from those messages
grep -i "competitor\|strengths\|weaknesses" /root/DATA/products/ContentForce.json | grep -o '"userId":"[^"]*"' | cut -d'"' -f4

Finding Shared URLs/Demo URLs

Pattern: Search for URLs, links, and demo references

# Find all URLs in messages
grep -o 'https*://[^" |<]*' /root/DATA/products/ContentForce.json | sort -u

# Find demo-related URLs specifically
grep -i "demo\|url\|link" /root/DATA/products/ContentForce.json | grep -o 'https*://[^" |<]*'

Deduplication and Formatting

# Remove duplicates and sort
grep -o 'eid_[a-f0-9]*' file.json | sort -u

# Count results
grep -o 'eid_[a-f0-9]*' file.json | sort -u | wc -l

Question Answering Workflow

Parse the question - Identify key terms (report names, keywords, types)
Search the product JSON - Use grep to find relevant messages
Extract entities - Get userIds, URLs, or other requested information
Deduplicate - Ensure unique results using sort -u
Format - Convert to the required output format (list of IDs/URLs)

Product Files Location

Main data: /root/DATA/products/
Each product has a .json file (e.g., ContentForce.json)
Look for both primary product and competitor product files

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.