Examples

The easiest way to understand OME is to look at a real prompt-time recall. This example uses a goal-start prompt because it shows the core difference between a generic rule file and a conditional experience card.

Example: /goal Starts Full-Closure Delivery

Assume your active library contains a reviewed card for goal execution. In this example the card is named 创建目标时进入完整闭环交付模式. Its usage criteria say to use it when /goal, 创建目标, or 开干 starts a real execution task, and to ignore it for documentation examples, feature explanations, or business-goal discussion.

When the user sends:

创建目标，开干，把这个功能完整做完并自己验证

OME decomposes the prompt into a task envelope. It detects goal execution wording, explicit execution intent, and real validation wording. That is enough to list the goal card as a high-risk, must-use candidate. Candidate does not mean automatically used; the agent still checks the card's workflow meaning against the current task before reading and applying the full rule.

What Gets Mounted Into The Agent Prompt

The hook mounts a compact context block. The frame is English so Codex and Claude receive one stable instruction shape. The card content stays in the language stored on the card.

OME candidate experience cards. Matched does not mean used: apply a card only when its workflow meaning fits the current task; ignore unrelated or conflicting cards.
Final report: if you actually used any card, add one final line `**本次使用 N条 OME 经验卡：** ...` using only the `Final link if used` values for cards you applied; omit the line if none applied.
1. [high risk][must] 创建目标时进入完整闭环交付模式 (创建目标时进入完整闭环交付模式-40383753)
   Summary: 当用户用 /goal、创建目标或开干启动真实长任务时，常见误判是只建目标或做局部切片；应进入完整闭环交付，并排除文档示例、概念解释和业务目标讨论。
   Scope: global
   Use if: /goal 开始执行; 创建目标开干; 使用 goal 跑长任务
   Ignore if: 文档或案例里展示 /goal; 解释 goal 功能但不执行
   Matched by: task looks like a real goal-execution start
   Rule: ome experience show 创建目标时进入完整闭环交付模式-40383753 --section rule
   Final link if used: [创建目标时进入完整闭环交付模式](<~/.oh-my-experience/experiences/active/创建目标时进入完整闭环交付模式-40383753.md>)

The linked path is rendered from the user's own library. A global library uses the user's dataDir; a project library uses <project-root>/.oh-my-experience/.

What The Agent Reads Next

The mounted context intentionally stays short. If the lesson applies, the agent then fetches the rule body:

ome experience show 创建目标时进入完整闭环交付模式-40383753 --section rule

The rule says:

当用户说 `/goal`、`创建目标`、`使用 goal`、`开干`、`开始执行目标`、`跑长任务`，或要求把一批需求压进目标推进时，必须把这视为执行启动协议，而不是普通目标文案。默认执行规则如下：

1. 启动前先明确目标、范围、非范围、真实完成标准和逐项验收清单；如果目标被切得太小，先指出范围风险并把用户已确认的一批需求纳入同一目标。
2. 按完整计划系统推进，锚定 source of truth、story、roadmap、设计方案或用户原话；执行中不能方向漂移，不能只做第一个可见切片。
3. 所有计划内功能点都要完整闭环，不交半成品；禁止只做 happy path、UI 壳、接口半截、placeholder、fake route、隐藏测试入口、内存替代、假外部动作或 fallback 制造两套真相。
4. 实现时同步保证可维护性、可扩展性、稳健性和鲁棒性：模块按真实边界拆分，职责清楚，必要时清理直接相关的脏逻辑和死代码；不要为了假想未来做空泛抽象。
5. 验证必须走真实入口和真实用户路径；命令、功能、状态、文档或证据要逐项覆盖。工具成功、局部 smoke、代码写完都不能单独代表完成。
6. 对复杂或高风险目标，完成后主动做自检；必要时 dispatch 外部模型或 review 流程，重点检查方向是否漂移、功能是否完整、实际可用性、架构质量和可维护性。
7. 完成判断必须 fail closed：只要还有计划功能未做完、验收未通过、证据缺失、环境阻塞未说明或风险未交代，就不能把目标标记为 complete；应继续修复或明确 blocked。
8. 最终交付要面向用户说明体验变化、已验证证据、风险限制和待确认项。

The result is not "always add more rules." The agent sees this as a candidate only when the current prompt looks like goal execution. If it actually uses the card, the final response can include **本次使用 1条 OME 经验卡：** .... If the user is only discussing business goals, OKRs, or asking what /goal means, the card is ignored and does not appear in the final usage line.

Try It Yourself

Use ome match to inspect the same path without sending a real agent prompt:

ome match "创建目标，开干，把这个功能完整做完并自己验证" --explain

Use --json when you want to inspect the task envelope, reasons, and rendered additionalContext programmatically:

ome match "创建目标，开干，把这个功能完整做完并自己验证" --explain --json