explainx / blog
Comprehensive guide to Anthropic's Knowledge Work Plugins for Claude Cowork and Claude Code. Learn how to install, customize, and build plugins for sales, engineering, product, marketing, and more.
Jul 9, 2026
Claude's new Reflect dashboard summarizes how you use AI across 1–12 months, scores you on Delegation/Description/Discernment/Diligence, and lets you set quiet hours — but only if Memory is on. explainx.ai breaks down the July 9 beta.
Jun 25, 2026
Claude Tag puts Claude inside Slack as a persistent, proactive team member. The @ClaudeDevs team shared how they use it to write 65% of their own code — including most of what built Claude Tag itself. Here are 15 specific use cases across engineering, ops, data, support, and documentation that show exactly what delegating to @Claude looks like in practice.
Jun 24, 2026
Claude Tag is Anthropic's evolution of Claude Code — now multiplayer, proactive, and woven into Slack. Tag @Claude in any channel and it breaks tasks into stages, works through them with the tools it has, and responds in the thread. One Claude per channel means teammates pick up exactly where you left off. It learns from channel history, takes initiative in ambient mode, and can plan work over hours or days. 65% of Anthropic's own product team's code now comes from their internal version.
TL;DR: Claude Knowledge Work Plugins transform Claude into a role-specific expert for your job function. With 15.8k GitHub stars and 11 official plugins covering sales, engineering, product, marketing, and more, these plugins integrate your tools, understand your workflows, and deliver finished professional work—not just suggestions.
Knowledge Work Plugins are role-specific extensions for Claude that bundle:
Repository: github.com/anthropics/knowledge-work-plugins (15.8k stars, 1.9k forks)
License: Apache 2.0 (free for commercial use)
Maintained by: Anthropic + 19 contributors
Built for: Claude Cowork and Claude Code
Old model:
New model:
Example:
Without plugins:
You: "Help me prepare for a sales call with Acme Corp"
Claude: "Here are some things you should research..."
You: *Goes to multiple tools, manually researches, takes notes*
With sales plugin:
You: /sales:call-prep Acme Corp
Claude: *Automatically pulls CRM data, researches company, analyzes past interactions, generates call agenda with talking points, identifies decision makers, suggests questions, creates follow-up template*
You: Review and use the finished prep doc
For: Everyone Helps with: Daily workflows, task management, personal organization
Key connectors:
What it does:
Example workflows:
Best for: Teams wanting consistent productivity across all roles
For: Sales reps, account executives, SDRs, sales managers
Key connectors:
What it does:
Slash commands:
/sales:call-prep [company] - Generate call preparation doc/sales:research-prospect [name] - Deep prospect research/sales:draft-outreach [context] - Personalized email sequences/sales:competitive-analysis [competitor] - Battlecard creation/sales:pipeline-review - Deal analysis and next stepsExample workflow: Call Prep
1. You: /sales:call-prep Acme Corp
2. Claude:
- Pulls CRM data (deal stage, contact info, past interactions)
- Researches company (news, funding, competitors)
- Analyzes past call transcripts (via Fireflies)
- Identifies key stakeholders and decision makers
- Generates agenda with talking points
- Suggests discovery questions
- Creates objection handling guide
- Drafts follow-up email template
3. Result: Complete call prep in 2 minutes vs. 30 minutes manually
ROI: Sales teams report 40-60% time savings on prep work
For: Support agents, success managers, technical support
Key connectors:
What it does:
Slash commands:
/support:triage [ticket] - Categorize and prioritize/support:draft-response [ticket] - Generate reply with context/support:escalate [ticket] - Package for tier 2/engineering/support:research-customer [account] - Pull full customer context/support:create-kb-article [topic] - Documentation from resolutionExample workflow: Ticket Response
1. New ticket arrives in Intercom: "Can't export data to CSV"
2. Claude (via plugin):
- Reads ticket context
- Checks customer account (tier, product version, history)
- Searches Guru KB for relevant articles
- Reviews similar past tickets
- Drafts personalized response with steps
- Suggests if escalation needed
3. Agent reviews, approves, sends
4. Result: 70% faster response time, consistent quality
ROI: 50-70% reduction in response time, 40% fewer escalations
For: Product managers, product owners, program managers
Key connectors:
What it does:
Slash commands:
/product:write-spec [feature] - Generate PRD/product:roadmap-plan [quarter] - Roadmap with prioritization/product:user-research-synthesis - Analyze feedback/product:stakeholder-update - Weekly/monthly updates/product:competitive-analysis [feature] - Market researchExample workflow: Write Product Spec
1. You: /product:write-spec "User authentication with SSO"
2. Claude:
- Reviews existing auth implementation (via Linear/Jira)
- Checks user feedback (via Intercom)
- Analyzes usage data (via Amplitude)
- Reviews competitive implementations
- Drafts complete PRD:
* Background and motivation
* User stories and acceptance criteria
* Technical requirements
* Success metrics
* Risk assessment
* Implementation phases
3. Result: First draft PRD in 10 minutes vs. 2-3 hours
ROI: 60% faster spec writing, better consistency
For: Content marketers, growth marketers, marketing managers
Key connectors:
What it does:
Slash commands:
/marketing:draft-content [type] [topic] - Generate content/marketing:plan-campaign [goal] - Multi-channel campaign plan/marketing:competitive-analysis [competitor] - Market positioning/marketing:performance-report [period] - Metrics analysis/marketing:seo-optimize [content] - SEO improvementsExample workflow: Campaign Planning
1. You: /marketing:plan-campaign "Q3 product launch"
2. Claude:
- Reviews past campaign performance (via HubSpot)
- Analyzes audience segments (via Amplitude)
- Researches competitive campaigns (via SimilarWeb)
- Checks SEO opportunities (via Ahrefs)
- Generates campaign plan:
* Target audience and personas
* Key messaging and positioning
* Channel mix and timeline
* Content calendar
* Success metrics and KPIs
* Budget allocation
* Risk mitigation
3. Result: Campaign plan in 15 minutes vs. 2 days
ROI: 3x faster campaign planning, better data integration
For: Corporate counsel, contract managers, compliance officers
Key connectors:
What it does:
Slash commands:
/legal:review-contract [file] - Analyze contract/legal:nda-triage [request] - NDA review workflow/legal:compliance-check [topic] - Regulatory guidance/legal:risk-assessment [issue] - Risk analysis/legal:draft-response [request] - Template-based draftingExample workflow: Contract Review
1. You: /legal:review-contract vendor-agreement.pdf
2. Claude:
- Extracts contract terms
- Identifies non-standard clauses
- Flags high-risk provisions
- Compares to company templates
- Checks compliance requirements
- Generates review memo:
* Key terms summary
* Risk assessment (high/medium/low)
* Red-flag clauses
* Negotiation recommendations
* Compliance notes
3. Result: Initial review in 5 minutes vs. 30-60 minutes
ROI: 75% faster initial review, consistent risk flagging
For: Controllers, financial analysts, accountants, CFO teams
Key connectors:
What it does:
Slash commands:
/finance:journal-entry [transaction] - Prepare entry/finance:reconciliation [account] - Reconcile account/finance:financial-statements [period] - Generate statements/finance:variance-analysis [period] - Analyze variances/finance:close-checklist [period] - Close process managementExample workflow: Month-End Close
1. You: /finance:close-checklist April 2026
2. Claude:
- Pulls GL data from data warehouse
- Identifies incomplete reconciliations
- Generates close checklist with status
- Drafts variance explanations for significant items
- Prepares preliminary financial statements
- Flags potential issues for review
- Creates audit trail documentation
3. Result: Close process organized in 20 minutes vs. 2-3 hours
ROI: 40% faster close process, fewer errors
For: Data analysts, analytics engineers, data scientists
Key connectors:
What it does:
Slash commands:
/data:write-query [question] - Generate SQL/data:analyze-results [data] - Statistical analysis/data:build-dashboard [metrics] - Dashboard design/data:validate-data [table] - Data quality checks/data:document-table [table] - Schema documentationExample workflow: Ad-Hoc Analysis
1. You: /data:write-query "Weekly active users by signup source, last 90 days"
2. Claude:
- Accesses data catalog (via Definite)
- Identifies relevant tables and columns
- Writes optimized SQL query
- Executes query (if permissions allow)
- Performs statistical analysis
- Creates visualization recommendations
- Generates insight summary
- Documents query for future use
3. Result: Analysis complete in 5 minutes vs. 30-45 minutes
ROI: 60% faster analysis, better documentation
For: Software engineers, DevOps, SREs, engineering managers
Key connectors:
What it does:
Slash commands:
/eng:code-review [PR] - Review pull request/eng:debug-issue [bug] - Investigation and root cause/eng:write-docs [feature] - Documentation generation/eng:tech-spec [feature] - Technical design doc/eng:incident-response [incident] - Incident managementExample workflow: Code Review
1. You: /eng:code-review #1234
2. Claude:
- Pulls PR from GitHub
- Reviews code changes
- Checks for common issues:
* Security vulnerabilities
* Performance problems
* Code style violations
* Test coverage gaps
- Compares to design specs (via Jira/Notion)
- Generates review comments
- Suggests improvements
3. Result: Initial review in 2 minutes vs. 15-30 minutes
ROI: 50% faster code reviews, consistent quality checks
For: Anyone who needs to find information across tools
Key connectors:
What it does:
Slash commands:
/search:find [query] - Search all tools/search:summarize [topic] - Aggregate information/search:related [document] - Find related contentExample workflow: Find Information
1. You: /search:find "Q4 pricing strategy"
2. Claude:
- Searches across:
* Slack conversations
* Notion pages
* Jira tickets
* Email threads
* Shared documents
- Ranks results by relevance
- Extracts key information
- Summarizes findings
- Identifies who to ask for more info
3. Result: Find information in 30 seconds vs. 10-15 minutes
ROI: 80% faster information retrieval
For: Life sciences researchers, biotech R&D, pharmaceutical scientists
Key connectors:
What it does:
Slash commands:
/bio:literature-search [topic] - Research synthesis/bio:target-analysis [gene/protein] - Target validation/bio:protocol-design [experiment] - Experimental protocol/bio:data-analysis [results] - Statistical analysisExample workflow: Target Validation
1. You: /bio:target-analysis BRCA1
2. Claude:
- Searches PubMed for latest research
- Checks Open Targets for druggability
- Reviews ChEMBL for existing compounds
- Analyzes ClinicalTrials.gov for ongoing studies
- Synthesizes:
* Function and mechanism
* Disease associations
* Druggability assessment
* Existing therapeutics
* Research gaps
* Clinical trial status
3. Result: Comprehensive target report in 10 minutes vs. 2-4 hours
ROI: 70% faster literature review, better synthesis
For: Plugin creators, team admins, customization needs
What it does:
Use cases:
Every plugin is just files—no code, no compilation, no complex build:
plugin-name/
├── .claude-plugin/
│ └── plugin.json # Manifest (metadata, commands)
├── .mcp.json # MCP server connections (tools)
├── commands/ # Slash commands (explicit actions)
│ └── command-name.md # Command definition and logic
└── skills/ # Domain knowledge (automatic)
└── skill-name.md # Skill instructions and expertise
What: Domain expertise Claude draws on automatically
Format: Markdown files with instructions and best practices
Example (simplified):
# Sales Call Preparation
## When to Use
Claude should use this skill when the user needs to prepare for a sales call.
## Process
1. Gather company information from CRM
2. Research recent news and funding
3. Review past interactions
4. Identify key stakeholders
5. Generate talking points
6. Create objection handling guide
## Best Practices
- Personalize for the specific industry
- Focus on customer pain points
- Include competitive positioning
- Suggest discovery questions
Skills activate automatically when Claude detects relevance—no explicit invocation needed.
What: User-triggered workflows via slash commands
Format: Markdown with command definition
Example:
# /sales:call-prep
## Description
Prepare comprehensive call preparation document
## Parameters
- company: Company name (required)
- contact: Contact name (optional)
## Workflow
1. Pull CRM data for company
2. Research company background
3. Review past communications
4. Identify decision makers
5. Generate call agenda
6. Create follow-up template
Commands are explicitly invoked by users when needed.
What: MCP server connections to external tools
Format: JSON configuration in .mcp.json
Example:
{
"mcpServers": {
"hubspot": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-hubspot"],
"env": {
"HUBSPOT_API_KEY": "${HUBSPOT_API_KEY}"
}
},
"slack": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-slack"],
"env": {
"SLACK_BOT_TOKEN": "${SLACK_BOT_TOKEN}",
"SLACK_TEAM_ID": "${SLACK_TEAM_ID}"
}
}
}
}
Connectors enable Claude to read/write data in your tools.
Step 1: Visit the plugin marketplace
Go to: claude.com/plugins
Step 2: Browse plugins by role
Step 3: Click "Install" on desired plugin
Step 4: Configure connectors (provide API keys)
Step 5: Start using slash commands and automatic skills
Step 1: Add the plugin marketplace
claude plugin marketplace add anthropics/knowledge-work-plugins
Step 2: Browse available plugins
claude plugin list --marketplace knowledge-work-plugins
Step 3: Install a specific plugin
# Install sales plugin
claude plugin install sales@knowledge-work-plugins
# Install multiple plugins
claude plugin install productivity@knowledge-work-plugins
claude plugin install engineering@knowledge-work-plugins
Step 4: Configure connectors
# Edit plugin configuration
claude plugin config sales
# Add API keys
export HUBSPOT_API_KEY=your-key
export SLACK_BOT_TOKEN=your-token
Step 5: Use in sessions
# Slash commands available automatically
> /sales:call-prep Acme Corp
# Skills activate when relevant
> Help me research this prospect: [email protected]
Generic plugins are useful, but customized plugins are game-changing:
Edit .mcp.json to use your preferred tools:
// Before: Generic sales plugin uses HubSpot
{
"mcpServers": {
"hubspot": { ... }
}
}
// After: Your company uses Salesforce
{
"mcpServers": {
"salesforce": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-salesforce"],
"env": {
"SALESFORCE_API_KEY": "${SALESFORCE_API_KEY}"
}
}
}
}
Edit skill files to include your specifics:
# Before: Generic product spec skill
## Process
1. Define problem
2. Propose solution
3. List requirements
# After: Your company's process
## Process
1. Define problem using Jobs-to-be-Done framework
2. Reference our design system (Figma: acme-design-system)
3. Check alignment with Q3 OKRs in Notion
4. Include security review checklist per SOC 2 requirements
5. Generate spec using our template: [link to template]
## Company Context
- Our ICP: Mid-market B2B SaaS companies (50-500 employees)
- Our values: Security-first, user-centric, data-driven
- Our stack: React, TypeScript, PostgreSQL, AWS
Modify commands to match your team's specific workflows:
# /product:write-spec [feature]
## Workflow (Customized for Acme Corp)
1. Check if feature aligns with product vision in Notion page "Product North Star"
2. Review user requests in Intercom tagged with feature name
3. Analyze usage data in Amplitude (dashboard: "Feature Usage")
4. Check engineering capacity in Linear milestone view
5. Review competitive landscape in our Ahrefs competitor tracking
6. Generate spec using Acme PRD template (Notion template ID: abc123)
7. Include:
- RICE prioritization score
- Security impact assessment (required for SOC 2)
- Mobile vs. web implementation decision
- Internationalization requirements (we support 12 languages)
- Accessibility compliance (WCAG 2.1 AA minimum)
Use the cowork-plugin-management plugin:
# Install plugin creator
claude plugin install cowork-plugin-management@knowledge-work-plugins
# Create new plugin
> /plugin:create-new real-estate
# Add skills
> /plugin:add-skill real-estate property-valuation
# Add commands
> /plugin:add-command real-estate /real-estate:comps-analysis
# Add connectors (MLS integration, Zillow, etc.)
> /plugin:add-connector real-estate mls-connector
Or build manually:
# 1. Create directory structure
mkdir -p my-plugin/.claude-plugin
mkdir -p my-plugin/skills
mkdir -p my-plugin/commands
# 2. Create manifest
cat > my-plugin/.claude-plugin/plugin.json <<EOF
{
"name": "my-plugin",
"version": "1.0.0",
"description": "Custom plugin for my team",
"commands": [
{
"name": "/my:command",
"description": "Description of command",
"path": "commands/my-command.md"
}
]
}
EOF
# 3. Add skills and commands (markdown files)
# 4. Add connectors (.mcp.json)
# 5. Install locally
claude plugin install ./my-plugin
Company: 50-person B2B SaaS company Plugin: Sales + Productivity Results:
Company: Fortune 500 tech company Plugin: Product Management + Enterprise Search Results:
Company: Mid-size corporate law firm Plugin: Legal (customized) Results:
Don't install all 11 plugins at once:
The magic happens when connectors work:
Don't try to perfect plugins on day one:
Create company-specific plugin versions:
Most people use 2-4 plugins together:
| Plugin | Primary Users | Key Tools | Commands | Connectors | Complexity |
|---|---|---|---|---|---|
| Productivity | Everyone | Slack, Notion, Asana | 8+ | 9 | Low |
| Sales | Sales teams | HubSpot, Slack, Clay | 12+ | 9 | Medium |
| Support | Support teams | Intercom, Guru, Jira | 10+ | 7 | Medium |
| Product | PMs | Linear, Figma, Amplitude | 15+ | 11 | High |
| Marketing | Marketers | HubSpot, Canva, Ahrefs | 12+ | 9 | Medium |
| Legal | Lawyers | Box, Jira, M365 | 8+ | 5 | Low |
| Finance | Finance teams | Snowflake, Excel | 10+ | 4 | High |
| Data | Analysts | Snowflake, Hex, Amplitude | 12+ | 6 | High |
| Engineering | Developers | GitHub, Jira, Slack | 15+ | 6 | High |
| Enterprise Search | Everyone | All connected tools | 5+ | 6 | Low |
| Bio-Research | Researchers | PubMed, Benchling | 10+ | 10 | High |
Combine plugins for complex workflows:
Example: Product Launch
1. /product:write-spec [feature] # Product Management plugin
2. /data:analyze-results [user research] # Data plugin
3. /eng:tech-spec [implementation] # Engineering plugin
4. /marketing:plan-campaign [launch] # Marketing plugin
5. /sales:battlecard [feature] # Sales plugin
Result: Complete launch plan across all functions
Some plugins work better together:
Build connectors for proprietary tools:
// Example: Custom CRM connector
import { McpServer } from '@modelcontextprotocol/sdk';
const server = new McpServer({
name: 'custom-crm',
version: '1.0.0',
});
server.tool({
name: 'get_account',
description: 'Retrieve account from custom CRM',
parameters: {
account_id: { type: 'string', required: true }
},
handler: async ({ account_id }) => {
// Call your CRM API
const account = await customCRM.getAccount(account_id);
return account;
}
});
Symptoms: Claude can't access tool data
Solutions:
Symptoms: Slash commands don't autocomplete
Solutions:
claude plugin list)Symptoms: Claude doesn't use skill knowledge
Solutions:
Symptoms: Commands take too long to execute
Solutions:
Based on GitHub activity and community requests:
Coming soon:
1. Improve existing plugins:
2. Create new plugins:
3. Build new connectors:
4. Share customizations:
# 1. Fork the repository
# 2. Create a feature branch
git checkout -b feature/new-plugin-or-improvement
# 3. Make changes (follow existing structure)
# 4. Test thoroughly
# 5. Submit pull request with:
# - Clear description
# - Use case explanation
# - Testing evidence
Knowledge Work Plugins transform Claude from a helpful chatbot into a specialist member of your team—one that knows your tools, understands your processes, and delivers finished work.
The progression:
Without Claude: Do everything manually ↓ With Claude: Get AI suggestions and guidance ↓ With Claude + Plugins: Set goals, get finished professional work
Start today:
With 15.8k stars and growing, the Knowledge Work Plugin ecosystem represents the future of work: AI that doesn't just assist—it collaborates, understands, and delivers.
Get started:
claude plugin marketplace add anthropics/knowledge-work-pluginsThe future of knowledge work isn't AI replacing humans—it's AI as a specialist teammate. Plugins make that future real, today.