tools-only Claude Skills - Skill-Agent Integration Implementation Plan
Version: 1.0 Created: November 5, 2025 Status: In Progress Milestone: Skill-Agent Integration v1.0
Executive Summary
This plan outlines the integration of specialized agents (cs-* prefix) with the existing 42 production-ready skills in the claude-code-skills repository. The implementation creates a seamless workflow where users can invoke domain-specific agents that automatically discover and execute the appropriate Python tools and reference knowledge bases.
Key Objectives:
- Create 5 foundational agents covering marketing, C-level advisory, and product management
- Implement simple installation system via interactive script
- Maintain 100% backwards compatibility with existing skill structure
- Prepare foundation for future Anthropic plugin marketplace distribution
Timeline: Days 1-4 (Phase 1-2) Future Work: Days 5-10 (Phase 3-7, including plugin creation)
Architecture Overview
Directory Structure
claude-code-skills/
├── agents/ # cs-* prefixed agents (NEW)
│ ├── marketing/
│ │ ├── cs-content-creator.md
│ │ └── cs-demand-gen-specialist.md
│ ├── c-level/
│ │ ├── cs-ceo-advisor.md
│ │ └── cs-cto-advisor.md
│ └── product/
│ └── cs-product-manager.md
├── commands/ # Slash commands (NEW)
├── standards/ # Standards library (NEW)
│ ├── communication/
│ ├── quality/
│ ├── git/
│ ├── documentation/
│ └── security/
├── templates/ # Templates (NEW)
├── marketing-skill/ # EXISTING - unchanged
├── c-level-advisor/ # EXISTING - unchanged
├── product-team/ # EXISTING - unchanged
├── project-management/ # EXISTING - unchanged
├── engineering-team/ # EXISTING - unchanged
├── ra-qm-team/ # EXISTING - unchanged
├── install.sh # Installation script (FUTURE)
├── uninstall.sh # Cleanup script (FUTURE)
├── INSTALL.md # Installation guide (FUTURE)
├── USAGE.md # Usage examples (FUTURE)
├── .claude/ # Dev-only workflows
└── .github/ # GitHub automation
Design Principles
- Non-invasive: Agents reference existing skills via relative paths
- Zero conflicts: cs-* prefix prevents collision with user's existing agents
- Backwards compatible: Existing skill folders remain untouched
- Plugin-ready: Structure designed for future marketplace distribution
- Simple installation: Single script with 3 Q&A questions (Phase 3)
Phase 1: Core Structure Setup (Day 1)
1.1 Create Root-Level Directories
Task: Set up foundational directory structure Owner: Implementation team Estimated Time: 30 minutes
Directories to create:
mkdir -p agents/marketing
mkdir -p agents/c-level
mkdir -p agents/product
mkdir -p commands
mkdir -p standards/communication
mkdir -p standards/quality
mkdir -p standards/git
mkdir -p standards/documentation
mkdir -p standards/security
mkdir -p templates
Success Criteria:
- All directories created in repository root
- Directory structure matches architecture diagram
- No conflicts with existing folders
GitHub Issue: #[TBD] - Create root-level directory structure
1.2 Port Core Standards from Factory
Task: Copy and adapt 5 essential standards from claude-code-skills-factory Owner: Documentation team Estimated Time: 2 hours
Standards to port:
communication-standards.md →
standards/communication/- Absolute honesty, zero fluff, pragmatic focus
- Critical analysis requirements
- Response protocol
- Prohibited responses
quality-standards.md →
standards/quality/- Code quality requirements
- Testing standards
- Review checklist
git-workflow-standards.md →
standards/git/- Conventional commits
- Branch naming
- PR requirements
- Commit message templates
documentation-standards.md →
standards/documentation/- Markdown formatting
- File naming conventions
- Structure requirements
- Living documentation principles
security-standards.md →
standards/security/- Secret detection
- Dependency scanning
- Security checklist
- Vulnerability reporting
Adaptation Notes:
- Replace factory-specific references with claude-skills context
- Update file paths to match new structure
- Reference existing skills (not rr-* agents)
- Maintain standard library focus (no external dependencies)
Success Criteria:
- All 5 standards files created and validated
- No broken links or factory-specific references
- Standards reference claude-skills architecture
- Files pass markdown linting
GitHub Issue: #[TBD] - Port core standards from factory
Phase 2: Agent Implementation (Days 2-4)
2.1 Create Marketing Agents (Day 2)
Task: Implement 2 marketing domain agents Owner: Marketing team Estimated Time: 4 hours
Agent 1: cs-content-creator
File: agents/marketing/cs-content-creator.md
Structure:
---
name: cs-content-creator
description: Create SEO-optimized marketing content with brand voice consistency
skills: content-creator
domain: marketing
model: sonnet
tools: [Read, Write, Bash, Grep, Glob]
---
# Content Creator Agent
## Purpose
Specialized agent for creating high-quality marketing content across multiple formats (blog posts, social media, email campaigns, video scripts). Integrates brand voice analysis, SEO optimization, and platform-specific best practices.
## Skill Integration
**Skill Location:** `../../marketing-skill/content-creator/`
### Python Tools
**Brand Voice Analyzer:**
```bash
python ../../marketing-skill/content-creator/scripts/brand_voice_analyzer.py input.txt
python ../../marketing-skill/content-creator/scripts/brand_voice_analyzer.py input.txt json
SEO Optimizer:
python ../../marketing-skill/content-creator/scripts/seo_optimizer.py article.md "primary keyword"
python ../../marketing-skill/content-creator/scripts/seo_optimizer.py article.md "primary keyword" "secondary,keywords"
Knowledge Bases
- Brand Guidelines:
../../marketing-skill/content-creator/references/brand_guidelines.md - Content Frameworks:
../../marketing-skill/content-creator/references/content_frameworks.md - Social Media Optimization:
../../marketing-skill/content-creator/references/social_media_optimization.md
Templates
- Content Calendar:
../../marketing-skill/content-creator/assets/content-calendar-template.md - Brand Voice Checklist:
../../marketing-skill/content-creator/assets/brand-voice-checklist.md
Workflows
Workflow 1: Create Blog Post
- Read brief from user
- Analyze existing brand voice samples using brand_voice_analyzer.py
- Reference content_frameworks.md for blog post structure
- Draft content following brand guidelines
- Run SEO analysis using seo_optimizer.py
- Refine based on SEO recommendations
- Output final content + SEO report
Workflow 2: Create Social Media Campaign
- Identify target platforms
- Reference social_media_optimization.md for platform best practices
- Draft platform-specific content (character limits, hashtags, CTAs)
- Validate brand voice consistency
- Generate content calendar using template
Workflow 3: Analyze Brand Voice
- Collect content samples from user
- Run brand_voice_analyzer.py on each sample
- Aggregate results to identify patterns
- Map to personality archetype (Expert, Friend, Innovator, Guide, Motivator)
- Create brand voice profile document
Integration Examples
Example 1: Blog post with SEO
# User provides brief
# Agent drafts article as draft.md
python ../../marketing-skill/content-creator/scripts/seo_optimizer.py draft.md "content marketing"
# Review SEO score and recommendations
# Refine draft based on suggestions
# Final output: optimized article
Example 2: Brand voice audit
# User provides 3 content samples
python ../../marketing-skill/content-creator/scripts/brand_voice_analyzer.py sample1.txt json > voice1.json
python ../../marketing-skill/content-creator/scripts/brand_voice_analyzer.py sample2.txt json > voice2.json
python ../../marketing-skill/content-creator/scripts/brand_voice_analyzer.py sample3.txt json > voice3.json
# Agent analyzes JSON outputs
# Identifies consistency patterns
# Recommends brand voice archetype
Success Metrics
- Content creation time reduced by 40%
- SEO scores improved by 30%+ on average
- Brand voice consistency across all outputs
- Platform-specific best practices followed
Related Agents
- cs-demand-gen-specialist (acquisition campaigns)
- cs-product-marketing (product launches)
References
- Skill Documentation:
../../marketing-skill/content-creator/SKILL.md - Standards:
../../standards/communication/communication-standards.md
#### Agent 2: cs-demand-gen-specialist
**File:** `agents/marketing/cs-demand-gen-specialist.md`
**Structure:** (Similar YAML frontmatter + content structure)
- Integrates with `../../marketing-skill/marketing-demand-acquisition/`
- Python tool: campaign_analyzer.py
- Workflows: Lead gen campaigns, conversion optimization, funnel analysis
**Success Criteria:**
- Both agents created with complete YAML frontmatter
- All relative paths validated (../../ references work)
- Python tool invocation examples tested
- Workflow sections comprehensive
- No broken links
**GitHub Issue:** #[TBD] - Create marketing agents (cs-content-creator, cs-demand-gen-specialist)
---
### 2.2 Create C-Level Advisory Agents (Day 3)
**Task:** Implement 2 C-level advisory agents
**Owner:** Strategy team
**Estimated Time:** 4 hours
#### Agent 3: cs-ceo-advisor
**File:** `agents/c-level/cs-ceo-advisor.md`
**Integration Points:**
- Skill: `../../c-level-advisor/ceo-advisor/`
- Python tools: strategic_framework_generator.py, scenario_planner.py, okr_tracker.py
- Knowledge bases: Strategic frameworks, decision templates, board reporting
**Workflows:**
- Strategic planning (3-year vision)
- Quarterly OKR setting
- Board deck preparation
- Scenario analysis (market shifts, competition)
#### Agent 4: cs-cto-advisor
**File:** `agents/c-level/cs-cto-advisor.md`
**Integration Points:**
- Skill: `../../c-level-advisor/cto-advisor/`
- Python tools: tech_stack_analyzer.py, architecture_auditor.py, team_velocity_tracker.py
- Knowledge bases: Tech stack decisions, architecture patterns, engineering metrics
**Workflows:**
- Technology roadmap planning
- Build vs buy analysis
- Technical debt assessment
- Engineering team scaling
**Success Criteria:**
- Both C-level agents created
- Strategic frameworks integrated
- Python tools documented with examples
- Workflow sections cover executive use cases
- References to CEO/CTO skill packages working
**GitHub Issue:** #[TBD] - Create C-level agents (cs-ceo-advisor, cs-cto-advisor)
---
### 2.3 Create Product Management Agent (Day 4)
**Task:** Implement product management agent
**Owner:** Product team
**Estimated Time:** 3 hours
#### Agent 5: cs-product-manager
**File:** `agents/product/cs-product-manager.md`
**Integration Points:**
- Skill: `../../product-team/product-manager-toolkit/`
- Python tools: rice_prioritizer.py, customer_interview_analyzer.py
- Knowledge bases: Product frameworks, roadmap templates, prioritization methods
**Workflows:**
- Feature prioritization (RICE framework)
- Customer interview analysis
- Product roadmap generation
- Quarterly planning
**Example Integration:**
```bash
# User uploads features.csv
python ../../product-team/product-manager-toolkit/scripts/rice_prioritizer.py features.csv --capacity 20 --output json
# Agent analyzes RICE scores
# Categorizes: Quick Wins, Big Bets, Maybes, Time Sinks
# Generates quarterly roadmap
Success Criteria:
- Product manager agent created
- RICE prioritization workflow functional
- Interview analysis integrated
- Roadmap generation examples documented
GitHub Issue: #[TBD] - Create product agent (cs-product-manager)
2.4 Create Agent Structure Template & Documentation (Day 4)
Task: Standardize agent creation process Owner: Documentation team Estimated Time: 2 hours
Deliverables:
Agent Template:
templates/agent-template.md- YAML frontmatter structure
- Required sections (Purpose, Skill Integration, Workflows, Success Metrics)
- Python tool invocation patterns
- Knowledge base reference format
- Related agents linking
Agent Creation Guide:
documentation/AGENT_CREATION_GUIDE.md- Step-by-step process
- Naming conventions (cs-* prefix)
- Relative path resolution
- Testing checklist
- Integration validation
Success Criteria:
- Template covers all required agent sections
- Creation guide is actionable (developers can create new agents)
- Examples from cs-content-creator referenced
- Template passes validation
GitHub Issue: #[TBD] - Create agent structure template and documentation
Future Phases (Days 5-10)
Phase 3: Installation System (Day 5)
- Create install.sh with 3 Q&A questions
- Create uninstall.sh
- Implement backwards compatibility detection
Phase 4: Documentation (Day 6)
- Update README.md with agent catalog
- Create INSTALL.md
- Create USAGE.md
- Update CLAUDE.md
Phase 5: Testing & Quality (Day 7)
- Validate YAML frontmatter
- Test relative path resolution
- Verify Python tool invocation
- Run quality gates
Phase 6-7: Plugin Creation (Days 8-10)
- Research Anthropic marketplace requirements
- Design plugin.yaml manifest
- Create plugin package structure
- Prepare submission assets (icon, screenshots, description)
- Submit to marketplace
Success Criteria
Phase 1-2 Completion Checklist
- All 4 root directories created (agents/, commands/, standards/, templates/)
- 5 core standards ported and validated
- 5 agents implemented and tested:
- cs-content-creator
- cs-demand-gen-specialist
- cs-ceo-advisor
- cs-cto-advisor
- cs-product-manager
- All relative paths (../../) resolve correctly
- Python tools execute successfully from agents
- Agent template created
- Documentation updated
- All tasks tracked in GitHub issues
- Issues linked to Milestone "Skill-Agent Integration v1.0"
- Issues synced to Project #9 board
Quality Gates
File Structure:
- All files use kebab-case naming
- Markdown files pass linting
- YAML frontmatter validates
Path Resolution:
- All
../../relative paths work - Python scripts execute from agent context
- Knowledge bases load correctly
- All
Documentation:
- No broken links
- All examples tested
- Workflow sections complete
Integration:
- Agents discover skills automatically
- Python tools execute with correct paths
- Templates load successfully
Risk Mitigation
Risk 1: Relative path resolution fails
- Probability: Medium
- Impact: High
- Mitigation: Test all paths in isolation before committing
- Fallback: Use absolute paths with environment variable
Risk 2: Agent naming conflicts with user's existing setup
- Probability: Low (cs-* prefix chosen to avoid conflicts)
- Impact: Medium
- Mitigation: Clear documentation, namespace everything
- Fallback: Allow users to customize prefix during installation
Risk 3: GitHub automation workflow issues
- Probability: Low (workflows already implemented and tested)
- Impact: Low
- Mitigation: Test issues creation before full rollout
- Fallback: Manual issue creation
Risk 4: Standards not applicable to all skills
- Probability: Medium
- Impact: Low
- Mitigation: Keep standards general, allow skill-specific overrides
- Fallback: Document exceptions in standards files
GitHub Issues Breakdown
Phase 1 Issues (Day 1)
Issue 1: Create root-level directory structure
- Labels:
type:enhancement,domain:agents,priority:high - Milestone: Skill-Agent Integration v1.0
- Checklist:
- Create agents/ with subdirectories (marketing/, c-level/, product/)
- Create commands/
- Create standards/ with subdirectories
- Create templates/
- Verify no conflicts with existing structure
- Update .gitignore if needed
Issue 2: Port core standards from factory
- Labels:
type:enhancement,domain:documentation,priority:high - Milestone: Skill-Agent Integration v1.0
- Checklist:
- Port communication-standards.md
- Port quality-standards.md
- Port git-workflow-standards.md
- Port documentation-standards.md
- Port security-standards.md
- Adapt all references to claude-skills context
- Validate markdown linting passes
Phase 2 Issues (Days 2-4)
Issue 3: Create marketing agents
- Labels:
type:enhancement,domain:agents,domain:marketing,priority:high - Milestone: Skill-Agent Integration v1.0
- Checklist:
- Create cs-content-creator.md with full structure
- Create cs-demand-gen-specialist.md
- Test relative paths to marketing-skill/
- Validate Python tool invocation
- Document workflows
- Add integration examples
Issue 4: Create C-level agents
- Labels:
type:enhancement,domain:agents,priority:high - Milestone: Skill-Agent Integration v1.0
- Checklist:
- Create cs-ceo-advisor.md
- Create cs-cto-advisor.md
- Test relative paths to c-level-advisor/
- Validate Python tool invocation
- Document strategic workflows
- Add executive use case examples
Issue 5: Create product agent
- Labels:
type:enhancement,domain:agents,domain:product,priority:high - Milestone: Skill-Agent Integration v1.0
- Checklist:
- Create cs-product-manager.md
- Test relative paths to product-team/
- Validate RICE prioritizer integration
- Validate interview analyzer integration
- Document roadmap workflows
Issue 6: Create agent structure template
- Labels:
type:documentation,domain:templates,priority:medium - Milestone: Skill-Agent Integration v1.0
- Checklist:
- Create templates/agent-template.md
- Create documentation/AGENT_CREATION_GUIDE.md
- Document YAML frontmatter requirements
- Document workflow section structure
- Add validation checklist
Metrics & KPIs
Implementation Metrics
- Time to complete Phase 1-2: 4 days (target)
- Number of agents created: 5
- Standards ported: 5
- GitHub issues created: 6-8
- Code review cycles: 1-2 per agent
Quality Metrics
- Path resolution success rate: 100%
- Python tool execution success rate: 100%
- Documentation coverage: 100% (all agents fully documented)
- Markdown linting pass rate: 100%
User Experience Metrics (Post Phase 3)
- Time to install: <5 minutes
- Agent discovery time: <1 minute
- Skill execution time: Same as direct Python invocation
- User satisfaction: TBD (collect feedback)
Next Steps After Phase 1-2
- Phase 3 Planning: Create detailed plan for installation system
- Community Feedback: Share Phase 1-2 implementation for early feedback
- Agent Expansion: Plan remaining 37 agents (project management, engineering, RA/QM)
- Plugin Research: Deep dive into Anthropic marketplace requirements
- Integration Testing: Comprehensive testing with real user workflows
Appendix A: Agent Naming Convention
Format: cs-{domain}-{role}
Examples:
- cs-content-creator (marketing domain)
- cs-demand-gen-specialist (marketing domain)
- cs-ceo-advisor (c-level domain)
- cs-product-manager (product domain)
Rules:
- Always use
cs-prefix (claude-skills) - Use kebab-case for multi-word roles
- Keep names concise (<30 characters)
- Match corresponding skill folder name when possible
Appendix B: Directory Tree (Complete)
claude-code-skills/
├── .claude/
│ └── commands/ # Development commands (git/, review.md, etc.)
├── .github/
│ ├── workflows/ # CI/CD automation
│ ├── AUTOMATION_SETUP.md
│ └── pull_request_template.md
├── agents/ # NEW: cs-* agents
│ ├── marketing/
│ │ ├── cs-content-creator.md
│ │ └── cs-demand-gen-specialist.md
│ ├── c-level/
│ │ ├── cs-ceo-advisor.md
│ │ └── cs-cto-advisor.md
│ └── product/
│ └── cs-product-manager.md
├── commands/ # NEW: User slash commands
├── standards/ # NEW: Standards library
│ ├── communication/
│ │ └── communication-standards.md
│ ├── quality/
│ │ └── quality-standards.md
│ ├── git/
│ │ └── git-workflow-standards.md
│ ├── documentation/
│ │ └── documentation-standards.md
│ └── security/
│ └── security-standards.md
├── templates/ # NEW: Templates
│ └── agent-template.md
├── documentation/
│ ├── implementation/
│ │ └── implementation-plan-november-2025.md # THIS FILE
│ ├── GIST_CONTENT.md
│ └── PYTHON_TOOLS_AUDIT.md
├── marketing-skill/ # EXISTING
├── c-level-advisor/ # EXISTING
├── product-team/ # EXISTING
├── project-management/ # EXISTING
├── engineering-team/ # EXISTING
├── ra-qm-team/ # EXISTING
├── CLAUDE.md
├── README.md
├── LICENSE
└── .gitignore
Document History
- v1.0 (2025-11-05): Initial implementation plan created for Phase 1-2
- Future versions will document Phase 3-7 planning and execution
End of Implementation Plan