A curated collection of AI agent personas, slash commands, skills, and MCP configurations for Claude Code and Cursor.
/plugin marketplace add Fohlio/ai-tools
/plugin install fohlio-ai-tools@fohlio-ai-tools
claude plugin marketplace update fohlio-ai-tools
claude plugin update fohlio-ai-tools@fohlio-ai-tools
Then restart Claude Code.
On macOS (case-insensitive filesystem), plugin marketplace add may fail with a rename error. Fix:
git clone [email protected]:Fohlio/ai-tools.git ~/.claude/plugins/marketplaces/fohlio-ai-toolsThen add to ~/.claude/plugins/known_marketplaces.json:
"fohlio-ai-tools": {
"source": { "source": "github", "repo": "Fohlio/ai-tools" },
"installLocation": "/Users/YOUR_USER/.claude/plugins/marketplaces/fohlio-ai-tools",
"lastUpdated": "2026-02-26T00:00:00.000Z"
}Then install:
claude plugin install fohlio-ai-tools@fohlio-ai-tools
Slash commands in commands/ orchestrate multi-agent workflows:
| Command | Description |
|---|---|
/fohlio-ai-tools:implement-feature |
Standard workflow: Plan → Implement → Test → Refactor → Verify |
/fohlio-ai-tools:implement-tdd-feature |
TDD workflow: Tests first → Implement → Refactor → Verify |
/fohlio-ai-tools:implement-bdd-feature |
BDD workflow: Gherkin scenarios → Tests → Implement → Verify |
/fohlio-ai-tools:test-code |
Analyze code, write automated tests (Playwright/Jest/Vitest) |
/fohlio-ai-tools:refactor-code |
Simplify and improve code readability and scalability |
/fohlio-ai-tools:prd-creation |
Create a comprehensive Product Requirements Document |
/fohlio-ai-tools:team-review |
Multi-agent review (Architect, UX, Sales, PM, Refactorer) |
/fohlio-ai-tools:business-review |
Product-marketing audit using 5C/SWOT framework |
Agent personas in agents/ with specialized expertise:
| Agent | Description |
|---|---|
code-architect |
Senior Full-Stack Architect (Clean Architecture, SOLID, DDD) |
code-tester |
QA Automation Engineer (Playwright, Jest, Vitest) |
code-refactorer |
Code simplification and maintainability expert |
ux-designer |
UX/UI Designer using Stitch and Figma |
project-manager |
Task decomposition, P0-P2 priorities, T-shirt sizing |
sales-marketer |
Growth marketing, CRO, and conversion optimization |
build-verificator |
Quality gate verification and completeness audits |
mobile-architect |
Flutter/cross-platform mobile specialist |
researcher |
Creative researcher: Web search, Context7, brainstorming |
Executable skills in skills/ for specialized tasks:
| Skill | Description |
|---|---|
feature-implementation |
Orchestrates Standard, TDD, or BDD workflows with quality gates |
prompt-engineering |
Research-backed prompt optimization for Claude, GPT, Gemini, etc. |
poc-hypothesis |
Quick technical validation with multiple hypotheses and self-testable BE endpoints |
bug-tracing |
Hypothesis-driven debugging with instrumented logging and user-assisted reproduction |
analyze-architecture |
Systematic 13-section architecture audit with scorecard output |
humanizer |
Remove AI writing patterns to sound more natural |
agent-browser |
Browser automation CLI for testing and interaction |
create-design-system |
Guides creation of design systems (goals, foundations, components) |
replicate |
Reverse-engineer a feature (architecture + user flow + design) into an integration PRD |
visual-poc |
Generate standalone HTML/CSS/JS visual POC from docs, AC, or PRD |
brainstorming |
Structured ideation using TRIZ, SCAMPER, and morphological analysis |
testbro |
AI-powered browser tests with TestBro CLI and AI logger |
Reference configurations in mcp/mcp.json:
| Server | Purpose |
|---|---|
| Context7 | Up-to-date library documentation |
| Playwright | Browser automation and testing |
| GitHub | Repository operations via Docker |
| Figma | Design context extraction |
| Stitch | AI-powered UI prototyping |
| Vercel | Deployment and hosting |
| Sentry | Error monitoring |
| DeepWiki | Documentation search |
| Agents Playbook | Agent workflow patterns |
export GITHUB_PERSONAL_ACCESS_TOKEN="ghp_..."
export OPENAI_API_KEY="sk-..."- Plan First — Always ask clarifying questions before coding
- Audit & Reuse — Search for existing patterns before creating new ones
- No Placeholders — Full implementation only, no TODOs or FIXMEs
- Context7 — Always use for up-to-date library documentation
- Verify — Never mark done without build verification
MIT