crm

name	CRM
description	Contact and opportunity management for PAI data layer. Manage contacts, track opportunities, follow up on relationships, and search the CRM. USE WHEN crm, add contact, new contact, crm:add-contact, pipeline, opportunities, follow-up, crm:follow-up, crm:pipeline, crm:search, search contacts, find contact, contact list.
argument-hint	["add-contact | pipeline | follow-up | search"]

Customization

Before executing, check for user customizations at: ~/.claude/PAI/USER/SKILLCUSTOMIZATIONS/CRM/

If this directory exists, load and apply any PREFERENCES.md, configurations, or resources found there. These override default behavior. If the directory does not exist, proceed with skill defaults.

🚨 MANDATORY: Voice Notification (REQUIRED BEFORE ANY ACTION)

You MUST send this notification BEFORE doing anything else when this skill is invoked.

1. Send voice notification:

   curl -s -X POST http://localhost:8888/notify \
     -H "Content-Type: application/json" \
     -d '{"message": "Running the WORKFLOWNAME workflow in the CRM skill to ACTION", "voice_id": "OqTGHgPzbq47nVmGUnK2", "voice_enabled": true}' \
     > /dev/null 2>&1 &
   

2. Output text notification:

   Running the **WorkflowName** workflow in the **CRM** skill...
   

This is not optional. Execute this curl command immediately upon skill invocation.

CRM Skill

Contact and opportunity management via the PAI data layer at $PAI_DATA_ROOT/CRM/. Contacts, opportunities, and organizations are markdown files with YAML frontmatter. All queries use ~/go/bin/yq with the --front-matter=extract flag.

Data Location

$PAI_DATA_ROOT is set via .env (default: ~/.claude/PAI/USER/DATA).

$PAI_DATA_ROOT/CRM/
├── contacts/          # contact-*.md files
├── opportunities/     # opp-*.md files
├── organizations/     # org-*.md files
└── context/           # rich context assets (analyses, PDFs, notes)
    ├── opp-{id}/      # assets scoped to an opportunity
    └── contact-{id}/  # assets scoped to a contact

Workflow Routing

Trigger	Workflow	Description
crm:add-contact, add contact, new contact	Workflows/AddContact.md	Create a new contact from natural language
crm:pipeline, pipeline, opportunities	Workflows/Pipeline.md	List opportunities by status
crm:follow-up, follow up, who to contact	Workflows/FollowUp.md	Find contacts not contacted in 30+ days
crm:search, find contact, search contacts	Workflows/Search.md	Full-text and frontmatter search

Data Writing Rules

• No emdashes: use colons, commas, or parentheses instead. Applies to all CRM data files.

Examples

• crm:add-contact Tibi from Sensorica, Holochain developer — Creates contact-tibi.md from natural language
• crm:pipeline — Lists all opportunities grouped by status
• crm:follow-up — Shows contacts where last_contact is 30+ days ago
• crm:search holochain — Returns all contacts/opportunities tagged with or mentioning holochain

Schemas

All entity types are defined in $PAI_DATA_ROOT/_schemas/. Templates are in $PAI_DATA_ROOT/_templates/.

Notes for the Homunculus

• One entity = one file. Never combine two people into one contact file.
• Frontmatter is the index. Body prose provides context; frontmatter enables querying.
• Always use --front-matter=extract with yq on .md files to avoid markdown body parse errors.
• id matches filename. contact-tibi.md has id: "contact-tibi".
• Context assets go in context/{entity-id}/ — not in the root context folder. Use opp-{id}/ for opportunity assets, contact-{id}/ for contact assets. Reference them in the parent entity's context_assets frontmatter field.
• Context assets can be any file type — markdown analyses, PDFs, email exports, notes. The context_assets field in the opportunity/contact frontmatter lists their paths.
• Invoices are managed by the ERP skill — use erp:invoice for invoice operations.