skill-system-insight

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Skill System Insight

技能系统洞察

This skill turns sessions into structured, explainable behavioral insights ("facets"), and uses those facets to evolve a user's Soul state over time.

The intent is pragmatic: collaborate better by learning the user's working style, not by inventing a persona.

该技能将会话转化为结构化、可解释的行为洞察（即“facet”），并利用这些facet逐步演化用户的Soul状态。

其核心目标务实：通过了解用户的工作方式而非构建虚拟角色，实现更高效的协作。

Architecture: Hybrid 3-Layer System

架构：混合三层系统

Layer 1: Base Profile (balanced.md)
  - Static skeleton: section format + safety/quality defaults.

Layer 2: Dual Matrix (soul-state)
  - Personality Matrix (slow/stable): openness, directness, autonomy, rigor, warmth
  - Emotion Matrix (faster baseline): patience, enthusiasm, caution, empathy

Layer 3: Synthesized Profile (user.md)
  - Periodically regenerated from Layer 1 + Layer 2 + accumulated facets.

Layer 2 is the ground truth. Layer 3 is a readable projection.

Layer 1: Base Profile (balanced.md)
  - Static skeleton: section format + safety/quality defaults.

Layer 2: Dual Matrix (soul-state)
  - Personality Matrix (slow/stable): openness, directness, autonomy, rigor, warmth
  - Emotion Matrix (faster baseline): patience, enthusiasm, caution, empathy

Layer 3: Synthesized Profile (user.md)
  - Periodically regenerated from Layer 1 + Layer 2 + accumulated facets.

Layer 2 是基准真值，Layer 3 是可读性更强的投影版本。

Data Model

数据模型

Facets (per-session extraction):
```
schema/facet.yaml
```
Soul state (dual matrix + counters/buffers):
```
schema/soul-state.yaml
```

Storage uses the Postgres

agent_memories

table:

Facets:

memory_type='episodic'

category='insight-facet'

Soul state:

memory_type='semantic'

category='soul-state'

Facets（每会话提取）：
```
schema/facet.yaml
```
Soul状态（双矩阵 + 计数器/缓冲器）：
```
schema/soul-state.yaml
```

存储使用Postgres的

agent_memories

表：

Facets：

memory_type='episodic'

，

category='insight-facet'

Soul状态：

memory_type='semantic'

，

category='soul-state'

Pipeline

流程

Trigger (manual or suggested) -> Extract Facet -> Update Matrix -> (optional) Synthesize Profile

References:

Facet extraction prompt:
```
prompts/facet-extraction.md
```
Soul synthesis prompt:
```
prompts/soul-synthesis.md
```
Extraction procedure:
```
scripts/extract-facets.md
```
Matrix update algorithm:
```
scripts/update-matrix.md
```
Profile regeneration procedure:
```
scripts/synthesize-profile.md
```

Trigger (manual or suggested) -> Extract Facet -> Update Matrix -> (optional) Synthesize Profile

参考资料：

Facet提取提示词：
```
prompts/facet-extraction.md
```
Soul合成提示词：
```
prompts/soul-synthesis.md
```
提取流程：
```
scripts/extract-facets.md
```
矩阵更新算法：
```
scripts/update-matrix.md
```
档案再生流程：
```
scripts/synthesize-profile.md
```

How To Trigger

触发方式

This is a manual workflow.

User can ask explicitly: "insight", "extract facets", "learn my preferences", "update my profile".
Router suggestion pattern (lightweight, non-pushy):
- "Want me to run an insight pass to learn from this session? (stores a facet + may update your matrix)"

When the user asks (or agrees), run:

```
scripts/extract-facets.md
```
```
scripts/update-matrix.md
```
If the synthesis trigger fires:
```
scripts/synthesize-profile.md
```

这是一个手动工作流。

用户可明确发起请求：“insight”、“extract facets”、“learn my preferences”、“update my profile”。
路由建议模式（轻量化、非强制）：
- “是否需要我运行洞察流程，从本次会话中学习？（将存储一个facet并可能更新您的矩阵）”

当用户发起请求（或同意建议）时，执行以下步骤：

```
scripts/extract-facets.md
```
```
scripts/update-matrix.md
```
若触发合成条件：
```
scripts/synthesize-profile.md
```

Constraints (Non-Negotiable)

约束条件（不可协商）

Transparency

透明度

Always explain what was learned and why.

Facets must contain evidence strings tied to concrete moments.
Matrix updates must add short context lines explaining each applied adjustment.

始终说明所学到的内容及原因。

Facet必须包含与具体时刻绑定的证据字符串。
矩阵更新必须添加简短的上下文说明，解释每一项调整的依据。

Rate limiting

频率限制

Max 3 facets per user per rolling 24 hours.

If over limit: do not store a new facet. Instead, write a short note to the user summarizing what you would have captured, and ask them to pick 1 session to record.

每个用户每24小时最多提取3个facet。

若超出限制：不存储新的facet，而是向用户发送简短说明，概述原本会记录的内容，并请用户选择1个会话进行记录。

Confidence threshold

置信度阈值

Do not change matrix values on a single observation.

Threshold: 3+ similar observations in the same direction.
Personality step size: +/- 0.05 per qualifying adjustment.
Emotion baseline step size: +/- 0.1 per qualifying adjustment.

Accumulation should be tracked (buffers) so the threshold is testable and explainable.

不得仅凭单次观察更改矩阵值。

阈值：同一方向的3次及以上相似观察。
人格矩阵步长：每次符合条件的调整±0.05。
情绪基准步长：每次符合条件的调整±0.1。

需跟踪积累情况（缓冲器），以便阈值可验证、可解释。

Where Layer 1 and Layer 3 Live

Layer 1和Layer 3的存储位置

Base profile (Layer 1):

skill/skills/skill-system-soul/profiles/balanced.md

Synthesized user profile (Layer 3):

skill/skills/skill-system-soul/profiles/<user>.md

The synthesis step should preserve the 6-section format used by

balanced.md

Identity
Decision Heuristics
Communication Style
Quality Bar
Tool Preferences
Anti-Patterns

基础档案（Layer 1）：

skill/skills/skill-system-soul/profiles/balanced.md

合成用户档案（Layer 3）：

skill/skills/skill-system-soul/profiles/<user>.md

合成步骤需保留

balanced.md

使用的6-section格式：

身份（Identity）
决策启发式（Decision Heuristics）
沟通风格（Communication Style）
质量标准（Quality Bar）
工具偏好（Tool Preferences）
反模式（Anti-Patterns）

Storage Pattern (agent_memories)

存储模式（agent_memories）

Example SQL templates (copy/paste and substitute values):

sql

-- Store a facet
SELECT store_memory(
  'episodic',
  'insight-facet',
  ARRAY['session:ses_xxx', 'user:arthu'],
  'Session Facet: <brief_summary>',
  '<full facet YAML as text>',
  '{"session_type": "...", "outcome": "..."}',
  'insight-agent',
  'ses_xxx',
  5.0
);

-- Store/update matrix state
SELECT store_memory(
  'semantic',
  'soul-state',
  ARRAY['user:arthu', 'matrix'],
  'Soul State: arthu',
  '<full soul-state YAML as text>',
  '{"total_insights": 0, "last_updated": "..."}',
  'insight-agent',
  NULL,
  9.0
);

-- Query recent facets
SELECT * FROM search_memories('insight-facet user:arthu', NULL, NULL, NULL, NULL, 0.0, 50);

示例SQL模板（复制粘贴并替换值）：

sql

-- Store a facet
SELECT store_memory(
  'episodic',
  'insight-facet',
  ARRAY['session:ses_xxx', 'user:arthu'],
  'Session Facet: <brief_summary>',
  '<full facet YAML as text>',
  '{"session_type": "...", "outcome": "..."}',
  'insight-agent',
  'ses_xxx',
  5.0
);

-- Store/update matrix state
SELECT store_memory(
  'semantic',
  'soul-state',
  ARRAY['user:arthu', 'matrix'],
  'Soul State: arthu',
  '<full soul-state YAML as text>',
  '{"total_insights": 0, "last_updated": "..."}',
  'insight-agent',
  NULL,
  9.0
);

-- Query recent facets
SELECT * FROM search_memories('insight-facet user:arthu', NULL, NULL, NULL, NULL, 0.0, 50);

Operational Notes

操作说明

Facet extraction should be completable in one pass. If you cannot justify an adjustment with concrete evidence, propose no adjustment.
Users may communicate in Chinese; treat that as a signal about comfort, not as a personality dimension.
Keep values clamped to [0.0, 1.0].

skill

{
  "schema_version": "2.0",
  "id": "skill-system-insight",
  "version": "1.0.0",
  "capabilities": ["insight-extract", "insight-matrix-update", "insight-synthesize"],
  "effects": ["db.read", "db.write", "fs.write"],
  "operations": {
    "extract-facets": {
      "description": "Extract a per-session facet from transcript. Rate limited to 3/24h per user.",
      "input": {
        "session_id": { "type": "string", "required": true, "description": "Session to analyze" },
        "user": { "type": "string", "required": true, "description": "User handle" }
      },
      "output": {
        "description": "Facet YAML stored to agent_memories",
        "fields": { "status": "ok | error", "memory_id": "integer" }
      },
      "entrypoints": {
        "agent": "Follow scripts/extract-facets.md procedure (no executable script)"
      }
    },
    "update-matrix": {
      "description": "Update dual matrix from stored facet with confidence gating.",
      "input": {
        "user": { "type": "string", "required": true, "description": "User handle" }
      },
      "output": {
        "description": "Updated soul-state YAML stored to agent_memories",
        "fields": { "status": "ok | error", "values_changed": "boolean" }
      },
      "entrypoints": {
        "agent": "Follow scripts/update-matrix.md procedure"
      }
    },
    "synthesize-profile": {
      "description": "Regenerate Layer 3 Soul profile from matrix + recent facets.",
      "input": {
        "user": { "type": "string", "required": true, "description": "User handle" }
      },
      "output": {
        "description": "User profile written to skill-system-soul/profiles/<user>.md",
        "fields": { "status": "ok | error", "profile_path": "string" }
      },
      "entrypoints": {
        "agent": "Follow scripts/synthesize-profile.md procedure"
      }
    }
  },
  "stdout_contract": {
    "last_line_json": false,
    "note": "Agent-executed procedures; output is structured YAML stored to DB, not stdout."
  }
}

Facet提取应一次完成。若无法用具体证据证明调整的合理性，则不提议任何调整。
用户可能使用中文沟通；这应视为用户对语言舒适度的信号，而非人格维度。
数值需限制在[0.0, 1.0]范围内。

skill

{
  "schema_version": "2.0",
  "id": "skill-system-insight",
  "version": "1.0.0",
  "capabilities": ["insight-extract", "insight-matrix-update", "insight-synthesize"],
  "effects": ["db.read", "db.write", "fs.write"],
  "operations": {
    "extract-facets": {
      "description": "从会话记录中提取每会话facet。每个用户每24小时最多3次。",
      "input": {
        "session_id": { "type": "string", "required": true, "description": "待分析的会话ID" },
        "user": { "type": "string", "required": true, "description": "用户标识" }
      },
      "output": {
        "description": "Facet YAML存储至agent_memories",
        "fields": { "status": "ok | error", "memory_id": "integer" }
      },
      "entrypoints": {
        "agent": "遵循scripts/extract-facets.md流程（无可执行脚本）"
      }
    },
    "update-matrix": {
      "description": "基于存储的facet更新双矩阵，带有置信度校验。",
      "input": {
        "user": { "type": "string", "required": true, "description": "用户标识" }
      },
      "output": {
        "description": "更新后的soul-state YAML存储至agent_memories",
        "fields": { "status": "ok | error", "values_changed": "boolean" }
      },
      "entrypoints": {
        "agent": "遵循scripts/update-matrix.md流程"
      }
    },
    "synthesize-profile": {
      "description": "基于矩阵和近期facet重新生成Layer 3 Soul档案。",
      "input": {
        "user": { "type": "string", "required": true, "description": "用户标识" }
      },
      "output": {
        "description": "用户档案写入skill-system-soul/profiles/<user>.md",
        "fields": { "status": "ok | error", "profile_path": "string" }
      },
      "entrypoints": {
        "agent": "遵循scripts/synthesize-profile.md流程"
      }
    }
  },
  "stdout_contract": {
    "last_line_json": false,
    "note": "由Agent执行的流程；输出为存储至数据库的结构化YAML，而非标准输出。"
  }
}
```",