provider-config-validator

vanman2024 的头像作者 vanman2024

验证和调试Vercel AI SDK提供程序配置,包括API密钥、环境设置、模型兼容性和速率限制。在遇到提供程序错误、身份验证失败、API密钥问题、缺少环境变量、模型兼容性问题、速率限制错误或用户提及提供程序设置、配置调试或SDK连接问题时使用。

📖 中文导读:验证和调试Vercel AI SDK提供程序配置,包括API密钥、环境设置、模型兼容性和速率限制。在遇到提供程序错误、身份验证失败、API密钥问题、缺少环境变量、模型兼容性问题、速率限制错误或用户提及提供程序设置、配置调试或SDK连接问题时使用。 以下正文为该技能的原始 SKILL.md 内容(英文),上方安装命令可直接使用。

name: provider-config-validator description: Validate and debug Vercel AI SDK provider configurations including API keys, environment setup, model compatibility, and rate limiting. Use when encountering provider errors, authentication failures, API key issues, missing environment variables, model compatibility problems, rate limiting errors, or when user mentions provider setup, configuration debugging, or SDK connection issues. allowed-tools: Read, Grep, Glob, Bash

Provider Configuration Validator

Purpose: Autonomously validate, diagnose, and fix Vercel AI SDK provider configurations.

Activation Triggers:

  • API errors (401, 403, 429, 404)
  • Missing/invalid API keys
  • Package installation issues
  • Model compatibility errors
  • Environment variable problems
  • Import/connection failures

Key Resources:

  • scripts/validate-provider.sh - Full validation (packages, keys, env)
  • scripts/check-model-compatibility.sh - Validate model names
  • scripts/generate-fix.sh - Generate fixes for common issues
  • scripts/test-provider-connection.sh - Test real API connection
  • examples/troubleshooting-guide.md - Comprehensive troubleshooting (10+ scenarios)
  • templates/ - .env, .gitignore, error handler code templates

Diagnostic Workflow

1. Detect Issue Type

Identify error from symptoms:

  • 401/403: Invalid API key → Check key format and env var
  • 429: Rate limiting → Add retry logic
  • 404/model_not_found: Invalid model → Validate model name
  • Cannot find module: Missing package → Install provider SDK
  • Missing env var: No .env → Create .env file
  • Import error: Wrong syntax → Fix imports for framework

2. Run Validation Script

# Main validation - checks everything
./scripts/validate-provider.sh <provider>

# Examples:
./scripts/validate-provider.sh openai
./scripts/validate-provider.sh anthropic

Checks performed:

  • ✅ Provider package installed (correct version)
  • ✅ Core SDK (ai) installed
  • ✅ .env file exists
  • ✅ API key set with correct format
  • ✅ .env in .gitignore (security)

3. Validate Model Name (if applicable)

./scripts/check-model-compatibility.sh <provider> <model>

# Examples:
./scripts/check-model-compatibility.sh openai gpt-4o
./scripts/check-model-compatibility.sh anthropic claude-sonnet-4-5-20250929

Shows valid models if name is wrong, suggests closest matches.

4. Generate Fixes

./scripts/generate-fix.sh <issue-type> <provider>

# Issue types:
# - missing-api-key      → Creates .env with correct format
# - wrong-format         → Shows valid key format
# - missing-package      → Installs provider package
# - model-compatibility  → Lists valid models
# - rate-limiting        → Adds retry helper with exponential backoff
# - import-error         → Fixes import statements

5. Test Connection

# Verify API credentials work
./scripts/test-provider-connection.sh <provider>

Makes real API call with minimal tokens to verify setup.

Provider-Specific Info

OpenAI:

  • Package: @ai-sdk/openai
  • Env: OPENAI_API_KEY
  • Format: sk-proj-... or sk-...
  • Models: gpt-4o, gpt-4o-mini, gpt-4, gpt-3.5-turbo

Anthropic:

  • Package: @ai-sdk/anthropic
  • Env: ANTHROPIC_API_KEY
  • Format: sk-ant-api03-...
  • Models: claude-sonnet-4-5-20250929, claude-opus-4-20250514

Google:

  • Package: @ai-sdk/google
  • Env: GOOGLE_GENERATIVE_AI_API_KEY
  • Format: AIza...
  • Models: gemini-1.5-pro, gemini-1.5-flash

xAI:

  • Package: @ai-sdk/xai
  • Env: XAI_API_KEY
  • Format: xai-...
  • Models: grok-beta, grok-vision-beta

Common Fixes

Missing API Key

# Create .env with correct structure
./scripts/generate-fix.sh missing-api-key openai

# Then add actual key from provider dashboard

Wrong Model Name

# Check valid models
./scripts/check-model-compatibility.sh anthropic claude-v3-opus
# Shows: Did you mean? → claude-opus-4-20250514

Rate Limiting

# Generate retry helper
./scripts/generate-fix.sh rate-limiting openai
# Creates retryHelper.ts with exponential backoff

Package Not Installed

# Install provider package
./scripts/generate-fix.sh missing-package anthropic
# Runs: npm install ai @ai-sdk/anthropic

Resources

Scripts: All scripts in scripts/ directory are executable and documented in README.md

Templates: templates/ contains .env, .gitignore, and error-handler-template.ts

Examples: examples/troubleshooting-guide.md has detailed solutions for 10+ scenarios including CORS, streaming, TypeScript, and provider-specific issues


Supported Providers: OpenAI, Anthropic, Google, xAI, Groq, Mistral, Cohere, DeepSeek

Version: 1.0.0 SDK Compatibility: Vercel AI SDK 5+