Enterprise data retrieval

Skill cxcscmu/SkillLearnBench/skills/b4-skill-creator-claude-sonnet-4-6/enterprise-information-search/enterprise-data-retrieval

Retrieve and analyze information from enterprise JSON data files (Slack messages, employee records, product data). Use this skill whenever the user asks to find employee IDs, URLs, or other structured information from enterprise data directories containing JSON files. Handles large files by using grep/search to locate relevant content before reading.From its SKILL.md

Install

npx -y skills add cxcscmu/SkillLearnBench --skill enterprise-data-retrieval

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

SKILL.md

2.0 KB, 394 tokens by cl100k_base, as published. Nobody here has run it

Enterprise Data Retrieval

Overview

Enterprise data is stored under /root/DATA/ with two subdirectories:

metadata/ — employee records, customer data, salesforce team info
products/ — per-product JSON files containing Slack channel messages, threads, and reactions

Data Structure

Each product JSON file has the structure:

{
  "slack": [
    {
      "Channel": { "name": "...", "channelID": "..." },
      "Message": {
        "User": { "userId": "eid_XXXX", "timestamp": "...", "text": "...", "utterranceID": "..." },
        "Reactions": []
      },
      "ThreadReplies": [ ... ],
      "id": "..."
    }
  ]
}

Key Patterns

Finding employee IDs

Employee IDs follow the pattern eid_XXXXXXXX
Use grep to find relevant messages: grep -i "keyword" /root/DATA/products/ProductName.json
Authors of documents are typically mentioned with phrasing like "I've shared the [Document]" or "I've created..."
Reviewers are those who reply with feedback/suggestions

Finding URLs

URLs in Slack messages are formatted as <URL|display text> or just plain URLs
Use grep to extract: grep -oP '<https?://[^|>]+' file.json

Large file handling

Product JSON files can be large (>256KB); use grep instead of reading the full file
Use offset/limit parameters when reading specific sections

Workflow

Identify which product file to search: /root/DATA/products/<ProductName>.json
Use grep with relevant keywords to locate messages
Extract employee IDs and URLs from matching context
Cross-reference with metadata/employee.json if needed for name-to-ID mapping

What ships with it

Read from the repository

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