AgentSkillsCN

Copywriting

文案写作

SKILL.md

Copywriting Skill

Purpose: Generate personalized LinkedIn messages and emails at scale using Clay's AI Writing best practices. Create targeted outreach that combines research signals with proven messaging frameworks.

When to Use This Skill

Use this skill when:

  • Generating personalized cold outreach for prospects
  • Creating LinkedIn connection requests and follow-up messages
  • Writing cold emails for job applications or client acquisition
  • Scaling personalized messaging without losing the human touch
  • Following up with researched opportunities from Opportunity Intelligence

Core Framework: FETC

Clay's FETC framework transforms GTM from basic data management to personalized, data-driven outreach at scale.

StepTraditionalAI-Powered
FindDatabase queries, list buildingAI discovers prospects based on subtle signals and complex attribute combinations
EnrichBasic contact detailsDeep firmographics, behavioral data, hiring signals, tech stack
TransformData cleaningStructuring, summarizing, extracting insights
CreateExport to CRMGenerate personalized messaging, custom content, strategic insights

AI Snippets Approach

The Clay Way: Generate individual sentences or small chunks, not entire paragraphs.

Why Snippets Work

ApproachProblemSolution
Full paragraphsAI drifts off-topic, loses coherenceGenerate one element at a time
Generic templatesFeels robotic, low response ratesInsert AI snippets into templates
Manual personalizationDoesn't scaleAI snippets + merge tags

Snippet Types

code
DYNAMIC OPENING LINES
├── Reference recent achievement
├── Note shared connection
├── Comment on content they posted
└── Acknowledge company news

CONTEXTUAL VALUE PROPOSITIONS
├── Analyze likely pain points
├── Match to your capabilities
├── Address specific challenges
└── Tie to their industry context

PERSONALIZED SOCIAL PROOF
├── Relevant case studies
├── Similar company results
├── Industry-specific examples
└── Role-specific wins

Personalization Signals

Data Points to Reference

Signal CategoryExamplesPersonalization Angle
Content ActivityBlog posts, LinkedIn articles, podcasts"I read your piece on X and noticed..."
Company NewsFunding, product launch, expansion"Congrats on the Series B! Given the growth..."
Hiring PatternsOpen roles, team growth"Saw you're scaling the RevOps team..."
Tech StackTools mentioned in job posts"Noticed you're evaluating HubSpot..."
Pain Signals"Fast-paced", "wear many hats""Supporting a lean team requires..."
Career TrajectoryRecent promotion, new role"Stepping into the VP role means..."

Personalization Depth Levels

code
LEVEL 1 - BASIC (Low effort, low response)
└── First name + company name only

LEVEL 2 - CONTEXTUAL (Medium effort, moderate response)
└── Industry + role + one specific detail

LEVEL 3 - RESEARCHED (High effort, high response)
└── Multiple signals + insight + value connection

LEVEL 4 - INTIMATE (Highest effort, best response)
└── Mutual connections + shared interests + specific problem

LinkedIn Message Templates

Connection Request (< 300 characters)

Formula: Signal + Relevance + Soft Ask

code
TEMPLATE 1 - Content Reference
"Hi [First Name], your [article/post] on [topic] resonated—especially
[specific point]. I work on similar challenges at [context]. Would love
to connect and exchange ideas."

TEMPLATE 2 - Company Signal
"Hi [First Name], congrats on [company news]. [Company] solving [problem]
is impressive. I help [similar companies] with [relevant value]. Happy to
connect?"

TEMPLATE 3 - Mutual Connection
"Hi [First Name], [Mutual Connection] mentioned your work on [initiative].
I'm doing similar work at [context]. Would be great to connect and compare
notes."

TEMPLATE 4 - Role Relevance
"Hi [First Name], fellow [role type] here—noticed you're building
[function] at [Company]. Currently doing the same at [context]. Always
learning from peers. Connect?"

Follow-Up Message

Formula: Acknowledgment + Value Add + Clear Ask

code
TEMPLATE - After Connection
"Thanks for connecting, [First Name]!

Digging into [Company]'s work on [specific initiative]—really smart
approach to [problem].

Based on what I'm seeing, thought you might find [specific resource/
insight] useful. [Brief explanation of why].

Would a quick chat make sense to compare notes on [specific topic]?"

Cold Email Templates

Structure: Problem-Agitate-Solve (PAS)

code
SUBJECT LINE
├── Personalized: "[First Name], quick thought on [specific thing]"
├── Curiosity: "[Company] + [your value] = ?"
├── Value-first: "Saved [similar company] X hours on [problem]"
└── Direct: "[Mutual] suggested I reach out"

OPENING (1-2 sentences)
├── Observation about their company
├── Reference to signal/trigger
├── Connection to pain point
└── NOT: "My name is... I work at..."

BODY (2-3 sentences)
├── Agitate the problem
├── Your relevant experience
├── Specific value you provide
└── Quick win offer

CTA (1 sentence)
├── Clear, low-friction ask
├── Give them an out
├── Time-specific options
└── "Worth a 15-min call, or should I share more context first?"

Email Templates by Purpose

Template 1: W2 Job Application

code
Subject: [Role Title] + quick-win idea

Hi [Hiring Manager Name],

Saw the [Role] posting and something jumped out—[specific requirement
or challenge mentioned]. I've solved exactly this at [Previous Company]
where I [specific accomplishment with metrics].

Looking at [Company]'s [current challenge/growth], I put together a
quick [teardown/audit/framework] showing [specific insight]. Attached
for your review—no strings attached.

If it resonates, would love to chat about how I could help [specific
outcome]. If not, hope it's useful regardless.

Best,
[Name]

Template 2: Freelance/Agency Pitch

code
Subject: [Specific problem I noticed] at [Company]

Hi [Name],

Noticed [specific signal—hiring, tech stack, growth indicator].
Companies at this stage often struggle with [common pain point].

I've helped [similar companies] solve this by [approach], typically
seeing [specific result with metrics].

Rather than a pitch deck, I put together a [quick deliverable: audit,
teardown, architecture diagram] for [Company]. [Link or attached]

Worth a 20-min call to walk through? Happy to share what I'm seeing
regardless of fit.

[Name]
ClayWorks of Art Agency

Template 3: Trojan Horse Project

code
Subject: Quick project idea for [specific pain point]

Hi [Name],

[Signal: funding, hiring, product launch] + [pain that comes with it]
is a recipe for [common problem].

Proposal: Let me run a 2-week [specific scoped project] covering:
- [Deliverable 1]
- [Deliverable 2]
- [Clear outcome with metrics]

Fixed scope, fixed price. You'll have [tangible output] whether we
work together again or not.

Interested?

[Name]

Prompt Engineering

Prompt Structure for AI Snippets

python
PROMPT_TEMPLATE = """
You are writing a {message_type} for cold outreach.

CONTEXT:
- Recipient: {name}, {title} at {company}
- Signal: {signal_detected}
- Pain point: {pain_point}
- Your offer: {value_proposition}

CONSTRAINTS:
- Maximum {char_limit} characters
- Must reference the specific signal
- No generic openings like "I hope this finds you well"
- Lead with value, not introduction
- {tone} tone

Generate {num_variations} variations.
"""

Prompt Best Practices

DoDon't
Tie each command to specific data pointsLeave scope too broad
Specify character/word limitsAsk for "paragraphs"
Include tone and style guidelinesAssume AI knows your brand
Request multiple variationsAccept first output
Provide examples of good outputGive vague instructions

Integration with Opportunity Intelligence

Workflow

code
OPPORTUNITY INTELLIGENCE
│
├── Account Research
│   ├── Company signals
│   ├── Pain points detected
│   └── Quick-win opportunities
│
├── Contact Research
│   ├── Decision maker identified
│   ├── LinkedIn activity
│   └── Career trajectory
│
└── COPYWRITING SKILL
    ├── Generate LinkedIn connection request
    ├── Generate follow-up message
    ├── Generate cold email
    └── Generate quick-win attachment pitch

Data Handoff

From Opportunity Intelligence, you'll receive:

  • company_summary: 2-3 sentence overview
  • pain_signals: List of detected challenges
  • outreach_angles: Personalization hooks
  • recommended_playbook: W2 or FREELANCE
  • quick_win_suggestions: Deliverables to offer
  • contact_info: Name, title, LinkedIn URL

API Configuration

GitHub Secrets

Secret NamePurpose
ANTHROPIC_API_KEYClaude API for message generation

Usage

python
import os

anthropic_key = os.environ.get("ANTHROPIC_API_KEY")

Quality Checklist

Before sending any AI-generated message:

  • Does it reference a specific signal, not generic?
  • Is the opening about them, not you?
  • Is there a clear value proposition?
  • Is the CTA low friction?
  • Does it sound human, not robotic?
  • Is it under appropriate length?
  • Have you verified the data used for personalization?

Common Pitfalls

PitfallProblemFix
Over-personalizationFeels stalker-like2-3 signals max
Data quality issuesWrong info damages credibilityVerify before sending
Generic templatesLow response ratesUse AI snippets
Too longWon't be readLinkedIn < 300, Email < 150 words
Weak CTANo clear next stepOne specific ask
All about youIgnoredLead with their problem

Resources

Included Files

FilePurpose
resources/templates.mdReady-to-use prompt templates and message examples
resources/writing_style_guide.mdVoice, tone, forbidden patterns, and quality standards
scripts/message_generator.pyPython script for automated message generation

Writing Style Guide

See resources/writing_style_guide.md for:

  • Forbidden patterns - Phrases to never use ("I hope this finds you well", "Just following up")
  • Voice and tone - Professional but not stuffy, direct but not blunt
  • Email templates - Cold outreach, follow-ups, meeting requests, bad news delivery
  • Client materials - Proposals, case studies, project updates
  • Internal communication - Slack best practices, status updates
  • Technical documentation - System docs, API docs, code comments
  • Style quick reference - Capitalization, numbers, formatting

Sources

This skill is based on Clay University's AI Writing at Scale best practices: