How long does the integration take?

Change one base URL in an OpenAI-compatible client, or install the Python or TypeScript SDK; the quickstart routes a first request in a few minutes.

Will this lock us into another vendor?

No. The router is Apache-2.0 open source, your provider keys stay yours, and the API is OpenAI-compatible, so leaving is a change of base URL.

What happens when a provider fails?

The router falls back to the next model your policy allows and records the fallback in the trace, so an outage costs a retry, not an incident.

QUICKSTART - Prompt - OpenSmartRoute

Name: QUICKSTART
Author: talent-factory

Prompt file imported from talent-factory/ai-project-example (.windsurf/workflows/skills/QUICKSTART.md). Fill in {{SKILL_NAME}} before use. Copyright stays with the author.

🚀 Skill Builder - Quick Start Guide

Build production-ready Claude Code Skills in minutes!

Fastest Path to Success

Step 1: Start Building

/skills:build-skill

That's it! The system will guide you through everything.

What Happens Next

Phase 1: Requirements (2-3 minutes)

skill-elicitation-agent asks you questions:

Example conversation:

Agent: "What specific task or workflow should this skill help with?"
You: "Help me write conventional commit messages"

Agent: "When should Claude automatically use this skill?"
You: "When I'm about to commit code or ask about commit messages"

Agent: "Does this need to execute code or just provide instructions?"
You: "Just instructions, no code needed"

... a few more questions ...

Agent: "Here's the specification I've created..."
[Shows detailed spec]

Agent: "Does this look good? Should I proceed to generation?"
You: "Yes, looks perfect!"

Phase 2: Generation (1-2 minutes)

skill-generator-agent builds your skill:

Creating skill structure...
✅ Created commit-helper/SKILL.md
✅ Added frontmatter with proper metadata
✅ Wrote clear instructions
✅ Included 3 examples
✅ Set up file structure

Location: ~/.claude/skills/commit-helper/

Phase 3: Validation (30 seconds)

skill-validator-agent tests everything:

Running 10 validation checks...
✅ YAML frontmatter valid
✅ Description clear and discoverable
✅ File structure correct
✅ No security issues

Score: 9/10 ⭐⭐⭐⭐⭐
Status: PRODUCTION READY

Phase 4: Documentation (1 minute)

skill-documenter-agent enhances docs:

Enhancing documentation...
✅ Added 5 comprehensive examples
✅ Created best practices section
✅ Added troubleshooting guide
✅ Included common pitfalls

Your skill is ready to use!

First-Time Examples

Example 1: Simple Instruction Skill (5 minutes)

What: Skill for commit messages

/skills:build-skill

Questions You'll Answer:

What format? → "Conventional Commits"
When to use? → "When committing code"
Any code needed? → "No, just instructions"

What You Get:

~/.claude/skills/commit-helper/
└── SKILL.md

How to Use:

"Help me write a commit message for my changes"

Example 2: Multi-File Skill with Scripts (10 minutes)

What: PDF form filling skill

/skills:build-skill

Questions You'll Answer:

What operations? → "Fill forms, extract fields"
Which libraries? → "pypdf, pdfplumber"
Execute code? → "Yes, need Python scripts"

What You Get:

~/.claude/skills/pdf-form-filler/
├── SKILL.md
├── FORMS.md
├── REFERENCE.md
└── scripts/
    ├── fill_form.py
    └── validate.py

How to Use:

"Fill out this PDF form with my data"

Example 3: Read-Only Safety Skill (7 minutes)

What: Security code analyzer

/skills:build-skill

Questions You'll Answer:

Analysis scope? → "OWASP Top 10"
Modify files? → "No, read-only"
Output format? → "Security report"

What You Get:

~/.claude/skills/security-analyzer/
├── SKILL.md (with tool restrictions)
├── PATTERNS.md
└── REFERENCE.md

How to Use:

"Analyze this code for security vulnerabilities"

Common Questions

Q: Where do skills get created?

A: By default in ~/.claude/skills/ (personal) or .claude/skills/ (project). The elicitation agent will ask which you prefer.

Q: Can I edit a skill after creation?

A: Yes! Just edit the files directly. Use the validation script to check:

.claude/commands/skills/scripts/validate-skill.sh ~/.claude/skills/my-skill

Q: How do I know if my skill will trigger?

A: Test it with:

.claude/commands/skills/scripts/test-skill-trigger.sh ~/.claude/skills/my-skill

Q: What if validation fails?

A: The validator will tell you exactly what to fix. The generator agent can help make the changes.

Q: Can I use templates directly?

A: Yes! Templates are in .claude/commands/skills/templates/. But using /skills:build-skill is recommended as it ensures quality.

Tips for Success

1. Be Specific in Elicitation

❌ Vague: "I need help with files"

✅ Specific: "I need to extract text and tables from PDF files, and also fill out PDF forms"

2. Provide Examples

When asked, share:

Example inputs
Desired outputs
Edge cases
Existing workflows

3. Review the Specification

The elicitation agent shows you the plan before building. Read it carefully!

4. Test Immediately

After creation, test with:

"Use [skill-name] to [example task]"

5. Iterate Based on Usage

Skills can be refined over time. Keep the specification document for reference.

Troubleshooting

Skill Not Loading

Check location:

ls ~/.claude/skills/
ls .claude/skills/

Check YAML:

head -20 ~/.claude/skills/my-skill/SKILL.md

Restart Claude Code to reload skills

Skill Not Triggering

Test triggers:

.claude/commands/skills/scripts/test-skill-trigger.sh ~/.claude/skills/my-skill

Use explicit invocation:

"Use the [skill-name] skill to [task]"

Validation Errors

Run validator:

.claude/commands/skills/scripts/validate-skill.sh ~/.claude/skills/my-skill

Fix issues listed in the report, then re-validate

Advanced Usage

Creating Project-Wide Skills

When asked about location, choose project instead of personal:

Location: .claude/skills/

Then commit to git:
git add .claude/skills/my-skill
git commit -m "Add my-skill for team"
git push

Team members get the skill automatically on pull.

Creating Plugin Skills

For distribution via plugins:

Create skill with /skills:build-skill
Move to plugin structure
Publish plugin to marketplace

See: https://docs.anthropic.com/claude-code/plugins

Using Templates Directly

For quick prototyping:

# Copy template
cp .claude/commands/skills/templates/simple-skill-template.md \
   ~/.claude/skills/my-skill/SKILL.md

# Replace placeholders
sed -i 's/{{SKILL_NAME}}/My Skill/g' ~/.claude/skills/my-skill/SKILL.md

# Validate
.claude/commands/skills/scripts/validate-skill.sh ~/.claude/skills/my-skill

But /skills:build-skill is recommended for production use.

What's Next?

Learn More

Full Documentation: See .claude/commands/skills/README.md
Agent Details: See .claude/agents/skill-builder/README.md
Official Docs: https://docs.anthropic.com/claude-code/skills

Build Your First Skill

Ready? Let's go:

/skills:build-skill

The agents will guide you through everything!

Share Your Skills

Once created:

Share with team via git
Distribute as plugins
Contribute to community repos

Happy skill building! Transform your workflows into reusable expertise. 🚀

Install QUICKSTART and it becomes one more candidate the router can pick - when it fits.

A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.

# 🚀 Skill Builder - Quick Start Guide Build production-ready Claude Code Skills in minutes! ## Fastest Path to Success ### Step 1: Start Building ```bash /skills:build-skill ``` That's it! The system will guide you through everything. ## What Happens Next ### Phase 1: Requirements (2-3 minutes) **skill-elicitation-agent** asks you questions: ``` Example conversation: Agent: "What specific task or workflow should this skill help with?" You: "Help me write conventional commit messages" Agent: "When should Claude automatically use this skill?" You: "When I'm about to commit code or ask about commit messages" Agent: "Does this need to execute code or just provide instructions?" You: "Just instructions, no code needed" ... a few more questions ... Agent: "Here's the specification I've created..." [Shows detailed spec] Agent: "Does this look good? Should I proceed to generation?" You: "Yes, looks perfect!" ``` ### Phase 2: Generation (1-2 minutes) **skill-generator-agent** builds your skill: ``` Creating skill structure... ✅ Created commit-helper/SKILL.md ✅ Added frontmatter with proper metadata ✅ Wrote clear instructions ✅ Included 3 examples ✅ Set up file structure Location: ~/.claude/skills/commit-helper/ ``` ### Phase 3: Validation (30 seconds) **skill-validator-agent** tests everything: ``` Running 10 validation checks... ✅ YAML frontmatter valid ✅ Description clear and discoverable ✅ File structure correct ✅ No security issues Score: 9/10 ⭐⭐⭐⭐⭐ Status: PRODUCTION READY ``` ### Phase 4: Documentation (1 minute) **skill-documenter-agent** enhances docs: ``` Enhancing documentation... ✅ Added 5 comprehensive examples ✅ Created best practices section ✅ Added troubleshooting guide ✅ Included common pitfalls Your skill is ready to use! ``` ## First-Time Examples ### Example 1: Simple Instruction Skill (5 minutes) **What**: Skill for commit messages ```bash /skills:build-skill ``` **Questions You'll Answer**: - What format? → "Conventional Commits" - When to use? → "When committing code" - Any code needed? → "No, just instructions" **What You Get**: ``` ~/.claude/skills/commit-helper/ └── SKILL.md ``` **How to Use**: ``` "Help me write a commit message for my changes" ``` ### Example 2: Multi-File Skill with Scripts (10 minutes) **What**: PDF form filling skill ```bash /skills:build-skill ``` **Questions You'll Answer**: - What operations? → "Fill forms, extract fields" - Which libraries? → "pypdf, pdfplumber" - Execute code? → "Yes, need Python scripts" **What You Get**: ``` ~/.claude/skills/pdf-form-filler/ ├── SKILL.md ├── FORMS.md ├── REFERENCE.md └── scripts/ ├── fill_form.py └── validate.py ``` **How to Use**: ``` "Fill out this PDF form with my data" ``` ### Example 3: Read-Only Safety Skill (7 minutes) **What**: Security code analyzer ```bash /skills:build-skill ``` **Questions You'll Answer**: - Analysis scope? → "OWASP Top 10" - Modify files? → "No, read-only" - Output format? → "Security report" **What You Get**: ``` ~/.claude/skills/security-analyzer/ ├── SKILL.md (with tool restrictions) ├── PATTERNS.md └── REFERENCE.md ``` **How to Use**: ``` "Analyze this code for security vulnerabilities" ``` ## Common Questions ### Q: Where do skills get created? **A**: By default in `~/.claude/skills/` (personal) or `.claude/skills/` (project). The elicitation agent will ask which you prefer. ### Q: Can I edit a skill after creation? **A**: Yes! Just edit the files directly. Use the validation script to check: ```bash .claude/commands/skills/scripts/validate-skill.sh ~/.claude/skills/my-skill ``` ### Q: How do I know if my skill will trigger? **A**: Test it with: ```bash .claude/commands/skills/scripts/test-skill-trigger.sh ~/.claude/skills/my-skill ``` ### Q: What if validation fails? **A**: The validator will tell you exactly what to fix. The generator agent can help make the changes. ### Q: Can I use templates directly? **A**: Yes! Templates are in `.claude/commands/skills/templates/`. But using `/skills:build-skill` is recommended as it ensures quality. ## Tips for Success ### 1. Be Specific in Elicitation ❌ **Vague**: "I need help with files" ✅ **Specific**: "I need to extract text and tables from PDF files, and also fill out PDF forms" ### 2. Provide Examples When asked, share: - Example inputs - Desired outputs - Edge cases - Existing workflows ### 3. Review the Specification The elicitation agent shows you the plan before building. Read it carefully! ### 4. Test Immediately After creation, test with: ``` "Use [skill-name] to [example task]" ``` ### 5. Iterate Based on Usage Skills can be refined over time. Keep the specification document for reference. ## Troubleshooting ### Skill Not Loading **Check location**: ```bash ls ~/.claude/skills/ ls .claude/skills/ ``` **Check YAML**: ```bash head -20 ~/.claude/skills/my-skill/SKILL.md ``` **Restart Claude Code** to reload skills ### Skill Not Triggering **Test triggers**: ```bash .claude/commands/skills/scripts/test-skill-trigger.sh ~/.claude/skills/my-skill ``` **Use explicit invocation**: ``` "Use the [skill-name] skill to [task]" ``` ### Validation Errors **Run validator**: ```bash .claude/commands/skills/scripts/validate-skill.sh ~/.claude/skills/my-skill ``` **Fix issues** listed in the report, then re-validate ## Advanced Usage ### Creating Project-Wide Skills When asked about location, choose **project** instead of personal: ``` Location: .claude/skills/ Then commit to git: git add .claude/skills/my-skill git commit -m "Add my-skill for team" git push ``` Team members get the skill automatically on pull. ### Creating Plugin Skills For distribution via plugins: 1. Create skill with `/skills:build-skill` 2. Move to plugin structure 3. Publish plugin to marketplace See: https://docs.anthropic.com/claude-code/plugins ### Using Templates Directly For quick prototyping: ```bash # Copy template cp .claude/commands/skills/templates/simple-skill-template.md \ ~/.claude/skills/my-skill/SKILL.md # Replace placeholders sed -i 's/{{SKILL_NAME}}/My Skill/g' ~/.claude/skills/my-skill/SKILL.md # Validate .claude/commands/skills/scripts/validate-skill.sh ~/.claude/skills/my-skill ``` But `/skills:build-skill` is recommended for production use. ## What's Next? ### Learn More - **Full Documentation**: See `.claude/commands/skills/README.md` - **Agent Details**: See `.claude/agents/skill-builder/README.md` - **Official Docs**: https://docs.anthropic.com/claude-code/skills ### Build Your First Skill Ready? Let's go: ```bash /skills:build-skill ``` The agents will guide you through everything! ### Share Your Skills Once created: - Share with team via git - Distribute as plugins - Contribute to community repos --- **Happy skill building! Transform your workflows into reusable expertise.** 🚀

{ "prompt": "# 🚀 Skill Builder - Quick Start Guide\n\nBuild production-ready Claude Code Skills in minutes!\n\n## Fastest Path to Success\n\n### Step 1: Start Building\n\n```bash\n/skills:build-skill\n```\n\nThat's it! The system will guide you through everything.\n\n## What Happens Next\n\n### Phase 1: Requirements (2-3 minutes)\n\n**skill-elicitation-agent** asks you questions:\n\n```\nExample conversation:\n\nAgent: \"What specific task or workflow should this skill help with?\"\nYou: \"Help me write conventional commit messages\"\n\nAgent: \"When should Claude automatically use this skill?\"\nYou: \"When I'm about to commit code or ask about commit messages\"\n\nAgent: \"Does this need to execute code or just provide instructions?\"\nYou: \"Just instructions, no code needed\"\n\n... a few more questions ...\n\nAgent: \"Here's the specification I've created...\"\n[Shows detailed spec]\n\nAgent: \"Does this look good? Should I proceed to generation?\"\nYou: \"Yes, looks perfect!\"\n```\n\n### Phase 2: Generation (1-2 minutes)\n\n**skill-generator-agent** builds your skill:\n\n```\nCreating skill structure...\n✅ Created commit-helper/SKILL.md\n✅ Added frontmatter with proper metadata\n✅ Wrote clear instructions\n✅ Included 3 examples\n✅ Set up file structure\n\nLocation: ~/.claude/skills/commit-helper/\n```\n\n### Phase 3: Validation (30 seconds)\n\n**skill-validator-agent** tests everything:\n\n```\nRunning 10 validation checks...\n✅ YAML frontmatter valid\n✅ Description clear and discoverable\n✅ File structure correct\n✅ No security issues\n\nScore: 9/10 ⭐⭐⭐⭐⭐\nStatus: PRODUCTION READY\n```\n\n### Phase 4: Documentation (1 minute)\n\n**skill-documenter-agent** enhances docs:\n\n```\nEnhancing documentation...\n✅ Added 5 comprehensive examples\n✅ Created best practices section\n✅ Added troubleshooting guide\n✅ Included common pitfalls\n\nYour skill is ready to use!\n```\n\n## First-Time Examples\n\n### Example 1: Simple Instruction Skill (5 minutes)\n\n**What**: Skill for commit messages\n\n```bash\n/skills:build-skill\n```\n\n**Questions You'll Answer**:\n- What format? → \"Conventional Commits\"\n- When to use? → \"When committing code\"\n- Any code needed? → \"No, just instructions\"\n\n**What You Get**:\n```\n~/.claude/skills/commit-helper/\n└── SKILL.md\n```\n\n**How to Use**:\n```\n\"Help me write a commit message for my changes\"\n```\n\n### Example 2: Multi-File Skill with Scripts (10 minutes)\n\n**What**: PDF form filling skill\n\n```bash\n/skills:build-skill\n```\n\n**Questions You'll Answer**:\n- What operations? → \"Fill forms, extract fields\"\n- Which libraries? → \"pypdf, pdfplumber\"\n- Execute code? → \"Yes, need Python scripts\"\n\n**What You Get**:\n```\n~/.claude/skills/pdf-form-filler/\n├── SKILL.md\n├── FORMS.md\n├── REFERENCE.md\n└── scripts/\n ├── fill_form.py\n └── validate.py\n```\n\n**How to Use**:\n```\n\"Fill out this PDF form with my data\"\n```\n\n### Example 3: Read-Only Safety Skill (7 minutes)\n\n**What**: Security code analyzer\n\n```bash\n/skills:build-skill\n```\n\n**Questions You'll Answer**:\n- Analysis scope? → \"OWASP Top 10\"\n- Modify files? → \"No, read-only\"\n- Output format? → \"Security report\"\n\n**What You Get**:\n```\n~/.claude/skills/security-analyzer/\n├── SKILL.md (with tool restrictions)\n├── PATTERNS.md\n└── REFERENCE.md\n```\n\n**How to Use**:\n```\n\"Analyze this code for security vulnerabilities\"\n```\n\n## Common Questions\n\n### Q: Where do skills get created?\n\n**A**: By default in `~/.claude/skills/` (personal) or `.claude/skills/` (project).\nThe elicitation agent will ask which you prefer.\n\n### Q: Can I edit a skill after creation?\n\n**A**: Yes! Just edit the files directly. Use the validation script to check:\n```bash\n.claude/commands/skills/scripts/validate-skill.sh ~/.claude/skills/my-skill\n```\n\n### Q: How do I know if my skill will trigger?\n\n**A**: Test it with:\n```bash\n.claude/commands/skills/scripts/test-skill-trigger.sh ~/.claude/skills/my-skill\n```\n\n### Q: What if validation fails?\n\n**A**: The validator will tell you exactly what to fix. The generator agent can help make the changes.\n\n### Q: Can I use templates directly?\n\n**A**: Yes! Templates are in `.claude/commands/skills/templates/`. But using `/skills:build-skill` is recommended as it ensures quality.\n\n## Tips for Success\n\n### 1. Be Specific in Elicitation\n\n❌ **Vague**: \"I need help with files\"\n\n✅ **Specific**: \"I need to extract text and tables from PDF files, and also fill out PDF forms\"\n\n### 2. Provide Examples\n\nWhen asked, share:\n- Example inputs\n- Desired outputs\n- Edge cases\n- Existing workflows\n\n### 3. Review the Specification\n\nThe elicitation agent shows you the plan before building. Read it carefully!\n\n### 4. Test Immediately\n\nAfter creation, test with:\n```\n\"Use [skill-name] to [example task]\"\n```\n\n### 5. Iterate Based on Usage\n\nSkills can be refined over time. Keep the specification document for reference.\n\n## Troubleshooting\n\n### Skill Not Loading\n\n**Check location**:\n```bash\nls ~/.claude/skills/\nls .claude/skills/\n```\n\n**Check YAML**:\n```bash\nhead -20 ~/.claude/skills/my-skill/SKILL.md\n```\n\n**Restart Claude Code** to reload skills\n\n### Skill Not Triggering\n\n**Test triggers**:\n```bash\n.claude/commands/skills/scripts/test-skill-trigger.sh ~/.claude/skills/my-skill\n```\n\n**Use explicit invocation**:\n```\n\"Use the [skill-name] skill to [task]\"\n```\n\n### Validation Errors\n\n**Run validator**:\n```bash\n.claude/commands/skills/scripts/validate-skill.sh ~/.claude/skills/my-skill\n```\n\n**Fix issues** listed in the report, then re-validate\n\n## Advanced Usage\n\n### Creating Project-Wide Skills\n\nWhen asked about location, choose **project** instead of personal:\n\n```\nLocation: .claude/skills/\n\nThen commit to git:\ngit add .claude/skills/my-skill\ngit commit -m \"Add my-skill for team\"\ngit push\n```\n\nTeam members get the skill automatically on pull.\n\n### Creating Plugin Skills\n\nFor distribution via plugins:\n\n1. Create skill with `/skills:build-skill`\n2. Move to plugin structure\n3. Publish plugin to marketplace\n\nSee: https://docs.anthropic.com/claude-code/plugins\n\n### Using Templates Directly\n\nFor quick prototyping:\n\n```bash\n# Copy template\ncp .claude/commands/skills/templates/simple-skill-template.md \\\n ~/.claude/skills/my-skill/SKILL.md\n\n# Replace placeholders\nsed -i 's/{{SKILL_NAME}}/My Skill/g' ~/.claude/skills/my-skill/SKILL.md\n\n# Validate\n.claude/commands/skills/scripts/validate-skill.sh ~/.claude/skills/my-skill\n```\n\nBut `/skills:build-skill` is recommended for production use.\n\n## What's Next?\n\n### Learn More\n\n- **Full Documentation**: See `.claude/commands/skills/README.md`\n- **Agent Details**: See `.claude/agents/skill-builder/README.md`\n- **Official Docs**: https://docs.anthropic.com/claude-code/skills\n\n### Build Your First Skill\n\nReady? Let's go:\n\n```bash\n/skills:build-skill\n```\n\nThe agents will guide you through everything!\n\n### Share Your Skills\n\nOnce created:\n- Share with team via git\n- Distribute as plugins\n- Contribute to community repos\n\n---\n\n**Happy skill building! Transform your workflows into reusable expertise.** 🚀", "variables": [ "SKILL_NAME" ], "notes": "Build production-ready Claude Code Skills in minutes!", "metadata": { "source": { "provider": "github-windsurf-workflows", "repository": "https://github.com/talent-factory/ai-project-example", "path": ".windsurf/workflows/skills/QUICKSTART.md", "ref": "0a3756e292fc9b2f2cc1c875a5001a0c28ad65f3", "url": "https://github.com/talent-factory/ai-project-example/blob/0a3756e292fc9b2f2cc1c875a5001a0c28ad65f3/.windsurf/workflows/skills/QUICKSTART.md", "key": "talent-factory/ai-project-example/.windsurf/workflows/skills/QUICKSTART.md" } } }

About

🚀 Skill Builder - Quick Start Guide

Fastest Path to Success

Step 1: Start Building

What Happens Next

Phase 1: Requirements (2-3 minutes)

Phase 2: Generation (1-2 minutes)

Phase 3: Validation (30 seconds)

Phase 4: Documentation (1 minute)

First-Time Examples

Example 1: Simple Instruction Skill (5 minutes)

Example 2: Multi-File Skill with Scripts (10 minutes)

Example 3: Read-Only Safety Skill (7 minutes)

Common Questions

Q: Where do skills get created?

Q: Can I edit a skill after creation?

Q: How do I know if my skill will trigger?

Q: What if validation fails?

Q: Can I use templates directly?

Tips for Success

1. Be Specific in Elicitation

2. Provide Examples

3. Review the Specification

4. Test Immediately

5. Iterate Based on Usage

Troubleshooting

Skill Not Loading

Skill Not Triggering

Validation Errors

Advanced Usage

Creating Project-Wide Skills

Creating Plugin Skills

Using Templates Directly

What's Next?

Learn More

Build Your First Skill

Share Your Skills

Install QUICKSTART and it becomes one more candidate the router can pick - when it fits.

What it costs to adopt - in engineering time.

One base URL

No vendor lock-in

Failover built in

Use it

Manifest

Reviews