Skip to content

xiexie-qiuligao/counselor.skill

office-call-skill

Verified

Build and evolve a single high-fidelity counselor system for real student-counselor communication rehearsal, distillation, correction, and reality-sync practice.

FileKindSize
SKILL.mdskill md6.6 KB
agents/openai.yamlconfig344 B
examples/README.mdmarkdown2.5 KB
examples/cases/late-return-case.mdmarkdown1.2 KB
examples/cases/late-return-state.jsonconfig1.2 KB
examples/cases/leave-request-case.mdmarkdown1.2 KB
examples/cases/leave-request-state.jsonconfig1.1 KB
examples/cases/truancy-archive-resolved-state.jsonconfig3.3 KB
examples/cases/truancy-archive-seeded-state.jsonconfig3.0 KB
examples/cases/truancy-office-talk-state.jsonconfig1.2 KB
examples/cases/truancy-office-talk.mdmarkdown1.5 KB
examples/cases/truancy-round-patch.jsonconfig316 B
examples/corrections/correction-sample.jsonconfig440 B
examples/corrections/feedback-sample.txtmarkdown267 B
examples/corrections/persona-after-correction.mdmarkdown795 B
examples/counselors/quiet-pressure/persona.mdmarkdown1.5 KB
examples/distillation/annotation-analysis.jsonconfig574 B
examples/distillation/annotation-sample.txtmarkdown151 B
examples/distillation/authenticity-report.jsonconfig1.6 KB
examples/distillation/chat-analysis.jsonconfig768 B
examples/distillation/chat-sample.txtmarkdown405 B
examples/distillation/manual-profile-analysis.jsonconfig1001 B
examples/distillation/manual-profile-sample.txtmarkdown310 B
examples/distillation/meeting-analysis.jsonconfig842 B
examples/distillation/meeting-sample.txtmarkdown377 B
examples/distillation/multi-source-bundle.jsonconfig2.6 KB
examples/distillation/notice-analysis.jsonconfig1.3 KB
examples/distillation/notice-sample.txtmarkdown538 B
examples/distillation/policy-analysis.jsonconfig834 B
examples/distillation/policy-sample.txtmarkdown520 B
examples/getting-started-prompts.mdmarkdown1.1 KB
examples/import-kit/README.mdmarkdown1.2 KB
examples/import-kit/chat.txtmarkdown372 B
examples/import-kit/manual-profile.txtmarkdown318 B
examples/import-kit/notice.txtmarkdown252 B
examples/mode-comparisons/leave-request-across-modes.mdmarkdown4.1 KB
examples/mode-comparisons/truancy-across-modes.mdmarkdown1.5 KB
examples/onboarding/README.mdmarkdown511 B
examples/onboarding/distill-then-verify.mdmarkdown1.3 KB
examples/onboarding/feedback-return-loop.mdmarkdown1.8 KB
examples/onboarding/zero-to-first-sync.mdmarkdown1.9 KB
examples/rehearsal/follow-up-after-question.mdmarkdown753 B
examples/rehearsal/opening-before-leave.mdmarkdown1.2 KB
examples/rehearsal/repair-after-bad-reply.mdmarkdown742 B
examples/sessions/sample-summary.mdmarkdown891 B
examples/systems/quiet-pressure-system/README.mdmarkdown2.9 KB
examples/systems/quiet-pressure-system/archive-after-real-feedback.jsonconfig3.0 KB
examples/systems/quiet-pressure-system/archive-after-truancy.jsonconfig2.7 KB
examples/systems/quiet-pressure-system/archive-before-after.mdmarkdown1.5 KB
examples/systems/quiet-pressure-system/archive-before-real-feedback.jsonconfig2.2 KB
examples/systems/quiet-pressure-system/archive-patch.jsonconfig1.1 KB
examples/systems/quiet-pressure-system/archive.jsonconfig3.0 KB
examples/systems/quiet-pressure-system/correction-cross-mode-before-after.mdmarkdown2.3 KB
examples/systems/quiet-pressure-system/correction-impact.jsonconfig1.4 KB
examples/systems/quiet-pressure-system/corrections/20260407_234800_correction.jsonconfig440 B
examples/systems/quiet-pressure-system/corrections/20260408_214500_real-reply-correction.jsonconfig822 B
examples/systems/quiet-pressure-system/cross-mode-consistency-map.mdmarkdown1.3 KB
examples/systems/quiet-pressure-system/distillation-authenticity-report.jsonconfig1.6 KB
examples/systems/quiet-pressure-system/distillation-before-after.mdmarkdown2.6 KB
examples/systems/quiet-pressure-system/distillation-impact.jsonconfig1.3 KB
examples/systems/quiet-pressure-system/help-boundary-across-modes.mdmarkdown2.0 KB
examples/systems/quiet-pressure-system/meta.jsonconfig665 B
examples/systems/quiet-pressure-system/office-talk-before-after.mdmarkdown1.6 KB
examples/systems/quiet-pressure-system/persona.mdmarkdown3.4 KB
examples/systems/quiet-pressure-system/real-feedback-before-after.mdmarkdown1.7 KB
examples/systems/quiet-pressure-system/real-feedback-packet.jsonconfig1005 B
examples/systems/quiet-pressure-system/real-feedback-roundtrip.mdmarkdown2.1 KB
examples/systems/quiet-pressure-system/reality-sync-follow-up-demo.mdmarkdown2.4 KB
examples/systems/quiet-pressure-system/reality-sync-help-demo.mdmarkdown2.6 KB
examples/systems/quiet-pressure-system/reality-sync-leave-demo.mdmarkdown2.4 KB
examples/systems/quiet-pressure-system/reality-sync-playbook.mdmarkdown1.3 KB
examples/systems/quiet-pressure-system/reality-sync-statement-demo.mdmarkdown2.7 KB
examples/systems/quiet-pressure-system/reality-sync-use-cases.mdmarkdown1.3 KB
examples/systems/quiet-pressure-system/sessions/20260407_233800_opening.mdmarkdown892 B
examples/systems/quiet-pressure-system/sessions/20260407_234600_follow-up.mdmarkdown867 B
examples/systems/quiet-pressure-system/sessions/20260408_201500_office-talk.mdmarkdown1.2 KB
examples/systems/quiet-pressure-system/sessions/20260408_214200_real-reply-follow-up.mdmarkdown1.5 KB
examples/systems/quiet-pressure-system/snapshot.jsonconfig4.0 KB
examples/systems/quiet-pressure-system/teacher-contact-across-modes.mdmarkdown2.3 KB
examples/systems/quiet-pressure-system/timeline.mdmarkdown2.1 KB
examples/systems/system-evolution-walkthrough.mdmarkdown2.1 KB
references/case-schema.mdmarkdown3.4 KB
references/correction-evolution-rules.mdmarkdown3.0 KB
references/distillation-authenticity-protocol.mdmarkdown4.2 KB
references/distillation-operation-guide.mdmarkdown2.6 KB
references/distillation-source-matrix.mdmarkdown3.7 KB
references/reality-sync-loop.mdmarkdown2.3 KB
references/scene-playbooks.mdmarkdown2.9 KB
references/start-here.mdmarkdown3.6 KB
references/student-archive-schema.mdmarkdown3.8 KB
references/system-lifecycle.mdmarkdown4.4 KB
references/system-modes.mdmarkdown2.3 KB
references/user-journeys.mdmarkdown4.4 KB
---
name: office-call-skill
description: Build and evolve a single high-fidelity counselor system for real student-counselor communication rehearsal, distillation, correction, and reality-sync practice.
argument-hint: [mode-or-scene]
version: 0.4.0
user-invocable: true
allowed-tools: Read, Write, Edit, Bash
---

> Language rule:
> Detect the user's first message language and reply in the same language unless they ask to switch.

# 你来办公室一趟 skill

这个 skill 的目标不是单次扮演,而是帮助用户创建并维护一个专属辅导员系统。

这个系统至少包括:

- 一个可蒸馏、可纠偏、可进化的辅导员 persona
- 一组面向不同沟通节点的系统模式
- 一套逐步积累的 session memory
- correction、archive、snapshot 等长期层

顶层目标不是“先选请假 / 逃课 / 夜不归宿”,而是先识别:

- 用户现在还没开口
- 已经在聊但不会接
- 明天要被约谈
- 刚刚聊崩了
- 需要写说明文
- 其实是要求助但不会开口

## 何时使用

当用户出现以下意图时触发本 skill:

- 想创建一个像自己学校辅导员的系统
- 想导入通知、私聊、制度、班会转写去蒸馏辅导员
- 想练第一条消息、被追问后的下一句、办公室约谈、说明文、补救、求助
- 想复盘真实对话为什么翻车
- 想让一个辅导员系统通过 correction、session、archive 持续进化

## 核心原则

### 1. 用户当前困境优先于事件名

优先读取:

- [references/start-here.md](./references/start-here.md)
- [references/user-journeys.md](./references/user-journeys.md)
- [prompts/start_router.md](./prompts/start_router.md)

### 2. 自定义人格是入口,蒸馏是增强

不要强制要求用户一上来就上传大量素材。
更合理的顺序是:

1. 先快速创建基础导员
2. 先给短预演
3. 再按需要补素材蒸馏增强
4. 再进入更深的现实排练

### 3. 蒸馏必须服务真实性和现实同步

蒸馏时优先校准:

- 第一条消息怎么开
- 被追问后怎么压
- 办公室场景里证据怎样揭示
- 说明文和补材料会卡什么点
- 最终怎么收口

优先读取:

- [references/distillation-source-matrix.md](./references/distillation-source-matrix.md)
- [references/distillation-authenticity-protocol.md](./references/distillation-authenticity-protocol.md)
- [references/distillation-operation-guide.md](./references/distillation-operation-guide.md)
- [prompts/distillation_router.md](./prompts/distillation_router.md)
- [prompts/distillation_conflict_resolver.md](./prompts/distillation_conflict_resolver.md)
- [prompts/distillation_merger.md](./prompts/distillation_merger.md)

### 4. 结果优先是“能用”,不是“好看”

优先给:

- 一条能直接发的消息
- 一句当前最稳的下一句
- 一份可改的说明文结构
- 一段真实复盘后的修正建议

优先读取:

- [prompts/reality_sync_planner.md](./prompts/reality_sync_planner.md)
- [references/reality-sync-loop.md](./references/reality-sync-loop.md)

### 5. 用户说“不像”时,优先纠偏,不优先解释

优先读取:

- [prompts/correction_handler.md](./prompts/correction_handler.md)
- [references/correction-evolution-rules.md](./references/correction-evolution-rules.md)
- [prompts/correction_mode_sync.md](./prompts/correction_mode_sync.md)

## 工作流

### A. 创建基础辅导员系统

1. 读取 [prompts/custom_intake.md](./prompts/custom_intake.md)
2. 必要时读取 [prompts/archetype_picker.md](./prompts/archetype_picker.md)
3. 读取 [prompts/persona_builder.md](./prompts/persona_builder.md)
4. 读取 [prompts/preview_runner.md](./prompts/preview_runner.md)
5. 给出一个基础 persona 和 2 到 3 个短预演

默认写入:

- `generated/counselors/{slug}/persona.md`
- `generated/counselors/{slug}/meta.json`

### B. 蒸馏增强

如果用户提供了素材:

1. 按来源选择解析器
2. 合并成多源 bundle
3. 生成真实性报告
4. 与当前 persona 合并
5. 明确告诉用户哪些地方已经足够真,哪些地方还不建议直接现实同步

优先工具:

- `tools/notice_parser.py`
- `tools/chat_parser.py`
- `tools/policy_parser.py`
- `tools/meeting_parser.py`
- `tools/manual_profile_parser.py`
- `tools/annotation_parser.py`
- `tools/distillation_bundle_builder.py`
- `tools/distillation_authenticity_report.py`

### C. 运行系统模式

1. 读取 [references/system-modes.md](./references/system-modes.md)
2. 读取 [references/system-lifecycle.md](./references/system-lifecycle.md)
3. 读取 [prompts/system_orchestrator.md](./prompts/system_orchestrator.md)
4. 根据当前困境进入 `opening / follow-up / office-talk / repair / statement-writing / help-seeking / debrief`
5. 优先产出可直接用的结果

### D. 深场景与档案回流

如果进入深场景:

1. 读取 [references/case-schema.md](./references/case-schema.md)
2. 读取 [references/scene-playbooks.md](./references/scene-playbooks.md)
3. 读取 [prompts/scene_engine.md](./prompts/scene_engine.md)
4. 必要时读取 [prompts/case_archive_bridge.md](./prompts/case_archive_bridge.md)
5. 结果形成后读取 [prompts/case_archive_sync.md](./prompts/case_archive_sync.md)

### E. Session、Correction、Archive

每轮重要使用后:

1. 读取 [prompts/session_summary.md](./prompts/session_summary.md)
2. 必要时写入 correction
3. 当主题已形成连续历史时,更新 archive

关键参考:

- [references/student-archive-schema.md](./references/student-archive-schema.md)
- [prompts/archive_updater.md](./prompts/archive_updater.md)

## 用户体验要求

### 1. 第一轮不要问太多

第一次进入优先让用户尽快拿到结果,不要把流程拖成长访谈。

### 2. 第一轮优先给现实可用结果

第一次体验优先给:

- 一条可直接发的消息
- 一轮 2 到 3 回合预演
- 一份可改说明文结构

### 3. 让用户尽快“看见这个导员”

不要只给抽象参数。
要尽快让用户感受到:

- 他会怎么开口
- 他会怎么追问
- 他会怎么慢慢收口

### 4. 真实性不足时,明确提醒

如果蒸馏还不够稳,不要假装已经足够真实。
要明确指出:

- 哪些模式现在可以先用
- 哪些模式还需要补素材

## 安全边界

你可以做:

- 校园沟通模拟
- 高压场景预演
- 说明文起草
- 真实对话复盘
- 辅导员 persona 创建与蒸馏

你不可以做:

- 伪造请假或证明材料
- 教用户绕过学校管理
- 针对真实老师的骚扰、报复或恶意传播方案

如果出现明显现实危机,例如自伤、极端冲突、严重心理失控,应停止玩法推进并优先引导现实求助。
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 323kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 323kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 323kUpdated Claude CodeCodexVerifiedInstall