Tencent Cloud Price Query

Problem Solved

When querying prices for various Tencent Cloud services (CVM, MySQL, Redis, Lighthouse, etc.), developers often need to assemble complex API parameters (region, instance type, disk type, billing mode) and navigate varying availability zone constraints. This skill accepts natural language or structured input, parses the configuration, and invokes local scripts/ to query prices, reducing the overhead of manually checking API documentation.

How It Works

• Multi-product Routing & Querying: Automatically matches input keywords to corresponding scripts (e.g., scripts/parse_input.py to scripts/query_trtc_price.py). Supports parsing quantities (e.g., "3 instances") and returning total prices.
• Spec & Billing Constraint Handling:
• CVM: Strictly follows instance type mapping (e.g., SA5.2XLARGE8) and validates zone availability in real-time via API to avoid hardcoding region limits.
• CDB MySQL: Identifies single-node, dual-node universal/economical, and three-node architectures. The economical type strictly enforces fixed CPU:Memory:Disk bindings.
• Redis: Supports standard/cluster architectures, parsing shard counts and replicas.
• Lighthouse: Calls DescribeBundles to fetch available packages, matches the closest BundleId, and then queries the price.
• Quote Generation: After querying, it can invoke scripts/generate_quote.py to generate an Excel quote, auto-filling actual specs, quantities, and durations, with optional fields for client and manager names.

Boundaries & Considerations

• Data Source Priority: TRTC, IM, Push, CDN, and COS default to local outputs/ JSON price files, falling back to online queries only if files are missing or corrupted; other products rely on real-time Tencent Cloud API calls.
• Account & Permissions: Querying APIs like CVM requires an activated service, real-name verification, and valid SecretId / SecretKey with appropriate API permissions.
• Spec Limitations: CDB memory must be a multiple of 1000, and Redis single-shard capacity must be a multiple of 1024. Some instance types may be sold out in specific regions, subject to real-time API validation.