customer-research
Multi-source research on a customer question or topic with source attribution. Use when a customer asks something you need to look up, investigating whether a bug has been reported before, checking what was previously told to a specific account, or gathering background before drafting a response.
Security Assessment
About customer-research
The customer-research skill performs multi-source research on a customer question, product topic, or account inquiry, synthesizing findings from all available sources with clear attribution and confidence scoring. It is used to look up answers for customers, check whether a bug was reported before, review what an account was previously told, or gather background before drafting a response.
It first parses the request type (customer question, issue investigation, account context, or topic research) and clarifies the goal and audience, then searches systematically across source tiers: official internal docs and knowledge bases, organizational context like CRM and support tickets, team communications, external web sources, and finally inference by analogy, cross-referencing rather than stopping at the first result. It synthesizes a structured brief leading with the answer and a confidence level, then key findings by source, context and nuance, cited sources, gaps, and recommended next steps, with explicit rules for confidence by source tier, handling contradictions, and when to escalate versus answer directly.
It is most useful for support and customer-facing teams who need well-sourced, confidence-rated answers and want to build institutional knowledge by capturing findings.
FAQ
When should I use this skill?
When you need to look up an answer to a customer question, investigate whether an issue was reported before, check an account's history, or gather background before drafting a response.
How does it search?
Systematically across source tiers (official internal docs, CRM and support, team chat and email, external web, and inference), cross-referencing rather than stopping at the first result.
What does the output look like?
A research brief that leads with the answer and a confidence level, then key findings by source, context and nuance, cited sources, gaps, and recommended next steps.
How is confidence determined?
By source tier, from high for official documentation down to low for inference, with explicit communication of the level and what drives it.
How does it handle conflicting sources?
It notes the contradiction, identifies the more authoritative or recent source, presents both perspectives, and for customer-facing answers uses the most conservative one until resolved.
Install customer-research
Quick Setup:
- Copy the skill folder to
.claude/skills/ - Claude will automatically detect and use the skill
Repository
anthropics/knowledge-work-plugins