SKILL.md
User Context Summarization
Turn enriched company and personal data into a structured seller context that other CRM skills use as input.
Inputs
The user provides one or more of:
- Their company name or domain
- Their LinkedIn profile URL
- Their Attio workspace context
- Manual overrides for any section
If not provided, attempt to infer from:
- The
.envfile (company domain patterns in email addresses) - The Extruct context file at
context/extruct_context.md(if it exists) - Ask the user
Workflow
Step 1: Gather company data
From Extruct context file (preferred)
Check if context/extruct_context.md exists in the repo. If so, read it for:
- Product description, value prop
- ICP definition
- Win cases and proof points
- Voice rules
From web (fallback)
If no local context exists, use web search to research the company.
Step 2: Gather personal data
From LinkedIn (if URL provided)
Use the Extruct Reverse Lookup table (via gettabledata on the Extruct MCP, using the EXTRUCTREVERSELOOKUPTABLEID from .env) or AnySite MCP to pull LinkedIn profile data.
From Attio (MCP)
Use search-records with object = "people" and the user's name or email to find their record. Extract:
- Job title, role
- Email signature patterns
- Communication style (from sent emails via Gmail MCP)
From Gmail (tone analysis)
Call gmailsearchmessages with from:me (limit 5) to sample the user's writing tone.
Step 3: Generate the context profile
Produce a structured document with these sections:
# Seller Context
## What I Sell
- **Product:** {one-line description}
- **Category:** {market category}
- **Key capabilities:** {3-5 bullet points}
- **Primary value prop:** {one sentence}
- **Differentiators:** {what makes it different from alternatives}
- **Proof points:** {notable customers, metrics, case studies}
## Who I Sell To
- **ICP:** {ideal customer profile — industry, size, stage}
- **Buyer personas:** {titles and roles that buy}
- **Pain points addressed:** {2-3 core problems solved}
- **Common objections:** {what prospects push back on}
- **Competitors they evaluate:** {who else they look at}
## Who Am I
- **Name:** {full name}
- **Role:** {job title}
- **Company:** {company name}
- **Tone:** {communication style — formal/casual, technical/accessible}
- **Signature style:** {how emails typically end}
- **Relationship approach:** {consultative, transactional, technical advisor}
Step 4: Save and confirm
Save the context profile to revops/seller_context.md.
Ask the user to review and correct any assumptions. Update the file with their edits.
This file is referenced by:
meeting-followup— for tone matchingmeeting-prep— for understanding what to pitchdeal-reengagement— for value prop and angle generationdeal-intelligence— for relationship framing
Environment
| Variable | Source | Purpose |
|---|---|---|
| Extruct | MCP (https://api.extruct.ai/mcp) |
Company and people enrichment |
| Attio | MCP (Attio connector) | CRM — people records |
| Gmail | MCP (claude.ai Gmail) |
Tone sampling |
Reference
- CRM enrichment context:
revops/CRM_ENRICHMENT.md