Agent Skills › QuixiAI/Hexis › crm-lookup

crm-lookup

GitHub

管理本地联系人数据库，支持搜索、查看、更新及合并重复记录。用于在对话中持久化人物信息，为会议准备等技能提供上下文，并处理心跳数据中的人物提及。

skills/installed/crm-lookup/SKILL.md QuixiAI/Hexis

Trigger Scenarios

查询特定人员详情其他技能需要联系人上下文对话中产生新的联系人信息需保存心跳处理涉及人物的数据发现并需合并重复联系人

Install

npx skills add QuixiAI/Hexis --skill crm-lookup -g -y

More Options

Non-standard path

npx skills add https://github.com/QuixiAI/Hexis/tree/main/skills/installed/crm-lookup -g -y

Use without installing

npx skills use QuixiAI/Hexis@crm-lookup

指定 Agent (Claude Code)

npx skills add QuixiAI/Hexis --skill crm-lookup -a claude-code -g -y

安装 repo 全部 skill

npx skills add QuixiAI/Hexis --all -g -y

预览 repo 内 skill

npx skills add QuixiAI/Hexis --list

SKILL.md

Frontmatter

{
    "name": "crm-lookup",
    "aliases": [
        "who",
        "contact",
        "contacts",
        "person",
        "people",
        "company",
        "account",
        "client",
        "customer",
        "owe",
        "deal"
    ],
    "category": "productivity",
    "contexts": [
        "heartbeat",
        "chat"
    ],
    "requires": {
        "tools": [
            "search_contacts",
            "get_contact"
        ]
    },
    "bound_tools": [
        "search_contacts",
        "get_contact",
        "update_contact",
        "merge_contacts",
        "create_contact",
        "ingest_contacts_email",
        "ingest_contacts_calendar"
    ],
    "description": "Search, view, update, and merge contacts in the local CRM"
}

CRM Contact Management

Manage the local contacts database -- search for people, view detailed profiles, update records with new information, and merge duplicates.

When to Use

When the user asks "who is [name]" or "find me [person]'s details"
When another skill (meeting-prep, email-digest) needs attendee or sender context
When new information about a contact surfaces during conversation and should be persisted
During heartbeats when processing ingested data that mentions people
When duplicates are suspected and need to be consolidated

Step-by-Step Methodology

Search first: Always start with search_contacts using the most specific identifier available (full name, email, company). Avoid overly broad queries that return too many results.
Disambiguate: If multiple contacts match, present the short list to the user (or, in heartbeat mode, pick the most relevant based on recency and relationship strength). Never silently pick the wrong person.
View details: Use get_contact to pull the full profile -- name, email, phone, company, role, tags, notes, and interaction history.
Enrich from memory: Cross-reference the contact with recall to find episodic memories of past interactions. This adds relational context that raw CRM fields may lack.
Update when warranted: If new information surfaces (new role, new company, corrected email), use update_contact to persist it. Always prefer updating an existing record over creating a new one.
Merge duplicates: If two records clearly represent the same person (same email, or same name + company), use merge_contacts to consolidate. The merge keeps the most complete data from both records.

Quality Guidelines

Treat contact data as sensitive. Never expose contact details to external services or tools without explicit user intent.
When updating contacts, preserve existing data. Do not overwrite a field with empty or less-specific information.
During merges, prefer the record with more complete data as the primary. Always log the merge as an episodic memory for audit.
If a search returns no results, say so clearly rather than guessing. Offer to create a new contact if appropriate.
Keep notes fields factual and professional. Store subjective relationship assessments in memories, not in the CRM record itself.