Agent Mentor Learn

参考来源

本课程的关键论断都以下列来源为依据。正文用 [^Sn] 引用，具体引文摘录在每条下方。

S1 — How we built our multi-agent research system — Anthropic Engineering

URL: https://www.anthropic.com/engineering/multi-agent-research-system

  • authority: authoritative-guide

Anthropic 多 Agent 研究系统的工程复盘，是本课程的核心依据：Agent 是有状态的、错误会复利；不能从头重启、要能从出错处恢复；可靠性靠「重试逻辑 + 定期检查点」这类确定性护栏与模型适应力的配合；部署更新时运行中的 Agent 可能处在流程任意一步。

关键引用：

"Agents can run for long periods of time, maintaining state across many tool calls." "Agents are stateful and errors compound." "Without effective mitigations, minor system failures can be catastrophic for agents." "When errors occur, we can't just restart from the beginning: restarts are expensive and frustrating for users." "Instead, we built systems that can resume from where the agent was when the errors occurred." "This means we need to durably execute code and handle errors along the way." "the adaptability of AI agents built on Claude with deterministic safeguards like retry logic and regular checkpoints" "letting the agent know when a tool is failing and letting it adapt works surprisingly well" "we use rainbow deployments to avoid disrupting running agents, by gradually shifting traffic from old to new versions" "whenever we deploy updates, agents might be anywhere in their process." "agents summarize completed work phases and store essential information in external memory before proceeding to new tasks"

S2 — Checkpointing — Claude Code Docs

URL: https://code.claude.com/docs/en/checkpointing

  • authority: official-docs

Claude Code 检查点机制的官方文档，提供「检查点/回退/分叉」在真实产品里的完整形态：每条用户消息自动建检查点、随会话保存、可分别恢复对话或代码、bash 改动不被追踪、不能替代版本控制。本课把它当作产品级对照，而非要教的工具。

关键引用：

"checkpointing automatically captures the state of your code before each user prompt." "Every user prompt creates a new checkpoint" "Claude Code saves checkpoints with the conversation, so you can still run /rewind after you resume a session" "Restore code and conversation: revert both code and conversation to that point" "Restore conversation: rewind to that message while keeping current code" "Restore code: revert file changes while keeping the conversation" "Exploring alternatives: try different implementation approaches without losing your starting point" "Recovering from mistakes: quickly undo changes that introduced bugs or broke functionality" "Checkpointing does not track files modified by bash commands." "Only direct file edits made through Claude's file editing tools are tracked." "Checkpoints are designed for quick, session-level recovery." "continue using version control, such as Git, for commits, branches, and long-term history." "To branch off and try a different approach while preserving the original session intact, use /branch or claude --continue --fork-session"

S3 — Building Effective AI Agents — Anthropic Engineering

URL: https://www.anthropic.com/engineering/building-effective-agents

  • authority: authoritative-guide

Anthropic 关于 Agent 模式的权威文章。本课用它支撑：Agent 可在检查点暂停等人反馈、自主性带来更高成本与误差累积、模型可能连跑很多轮而信任有限、以及「只在复杂度确实改善结果时才考虑增加」的分寸原则。

关键引用：

"Agents can then pause for human feedback at checkpoints or when encountering blockers." "The autonomous nature of agents means higher costs, and the potential for compounding errors." "The LLM will potentially operate for many turns, and you must have some level of trust in its decision-making." "it's also common to include stopping conditions (such as a maximum number of iterations) to maintain control." "you should consider adding complexity only when it demonstrably improves outcomes."

S4 — Handle tool calls — Claude API

URL: https://platform.claude.com/docs/en/agents-and-tools/tool-use/handle-tool-calls

  • authority: official-docs

给出 tool_use 与 tool_result 内容块的字段规范与配对规则：每个 tool_use 都必须有对应的 tool_result 集中回传。本课「断点恢复」一课依赖这条规则处理「崩溃发生在工具执行与结果落账之间」的悬空调用。

关键引用：

"id: A unique identifier for this particular tool use block. ... name: The name of the tool being used. input: An object containing the input being passed to the tool, conforming to the tool's input_schema. ... tool_use_id: The id of the tool use request this is a result for. ... is_error (optional): Set to true if the tool execution resulted in an error." "Whichever strategy you use, return one tool_result for each tool_use block, all together in the next user message. Match each result to its call with tool_use_id, and put every tool_result block before any text content in that message."

S5 — The 2026 Agent Engineering Roadmap — GitHub (codejunkie99/agent-roadmap-2026)

URL: https://github.com/codejunkie99/agent-roadmap-2026

  • authority: blog

以「harness 工程」为主线的社区开源路线图（AI 辅助撰写）。本课只取其框架性论点：持久化是 harness 组件之一，其职责是「每一步落检查点，以便恢复、回退、分叉」。注意：其中的百分比阈值、token 数、基准分数、薪资等具体数字一律不作为事实引用。

关键引用：

"the harness is the union of:" "persistence. Checkpoint state every node so you can resume, rewind, fork." "Same model, different harness, completely different result."