附录 D 参考文档列表
以下为本书引用的官方文档。各组括号注明核实日期。供应商文档经常变化，实际应用前请重新确认。
Anthropic (2026-10-20核实，*标记为10-21)
- Messages API reference
- Handling stop reasons
- Models overview
- Context windows
- Compaction on demand*
- Mid-conversation system messages*
- Token counting
- Effort
- Structured outputs
- Tool use overview
- Define tools
- Handle tool calls
- Parallel tool use
- Streaming messages
- Prompt caching
- Batch processing
- API errors
- Rate limits
- Define success criteria and build evaluations
- API and data retention
OpenAI (2026-10-20核实，*标记为10-21)
- Text generation
- Conversation state
- Compaction
- Reasoning models
- Structured outputs
- Function calling
- Latency optimization
- Streaming API responses
- Prompt caching
- Batch API
- Flex processing
- Fast mode
- Rate limits
- Error codes
- Background mode
- Webhooks
- Data controls in the OpenAI platform
- Safety best practices
- Evals
- openai-node SDK README*
Google (2026-10-20核实，*标记为10-21)
- Text generation
- What's new in Gemini 4 Argon
- Structured output
- Function calling
- Context caching
- Batch API
- Rate limits
- API errors
- Troubleshooting guide
- Interactions API overview*
- Batch API reference*
- Generating content API reference*
标准与公共技术文档（2026-10-20核实）
- OWASP — Top 10 for LLM Applications 2025
- OWASP — LLM01:2025 Prompt Injection
- OWASP — LLM05:2025 Improper Output Handling
- OWASP — LLM06:2025 Excessive Agency
- IETF — RFC 9110 HTTP Semantics §9.2.2 Idempotent Methods
- AWS Builders' Library — Timeouts, retries, and backoff with jitter
法律（2026-10-20核实）
- 韩国国家法令信息中心 — 个人信息保护法 （法律第21445号，2026年9月11日施行）。第2、3、23、26、28条之8及第29条