hermes-history-ingest

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Hermes History Ingest — Conversation & Memory Mining

Hermes历史记录导入——对话与记忆挖掘

You are extracting knowledge from the user's Hermes agent history and distilling it into the Obsidian wiki. Hermes stores both free-form memories and structured session transcripts — focus on durable knowledge, not operational telemetry.

This skill can be invoked directly or via the

wiki-history-ingest

router (

/wiki-history-ingest hermes

你需要从用户的Hermes Agent历史记录中提取知识，并提炼到Obsidian wiki中。Hermes会存储自由格式的记忆和结构化的会话记录——重点关注持久化知识，而非操作遥测数据。

此技能可直接调用，也可通过

wiki-history-ingest

路由调用（

/wiki-history-ingest hermes

）。

Before You Start

开始之前

Read

.env

to get

OBSIDIAN_VAULT_PATH

and

HERMES_HISTORY_PATH

(default to

~/.hermes

if unset)

Read
```
.manifest.json
```
at the vault root to check what has already been ingested
Read
```
index.md
```
at the vault root to understand what the wiki already contains

读取
```
.env
```
文件获取
```
OBSIDIAN_VAULT_PATH
```
和
```
HERMES_HISTORY_PATH
```
（若未设置则默认使用
```
~/.hermes
```
）
读取vault根目录下的
```
.manifest.json
```
文件，查看已导入的内容
读取vault根目录下的
```
index.md
```
文件，了解wiki已包含的内容

Ingest Modes

导入模式

Append Mode (default)

追加模式（默认）

Check

.manifest.json

for each source file. Only process:

Files not in the manifest (new memory files, new session logs)
Files whose modification time is newer than
```
ingested_at
```
in the manifest

Use this mode for regular syncs.

检查

.manifest.json

中的每个源文件。仅处理以下文件：

未在清单中的文件（新的记忆文件、新的会话日志）
修改时间晚于清单中
```
ingested_at
```
的文件

此模式用于常规同步。

Full Mode

全量模式

Process everything regardless of manifest. Use after

wiki-rebuild

or if the user explicitly asks for a full re-ingest.

无论清单如何，处理所有内容。在

wiki-rebuild

之后或用户明确要求全量重新导入时使用此模式。

Hermes Data Layout

Hermes数据结构

Hermes stores all local artifacts under

~/.hermes/

(or

$HERMES_HOME

for non-default profiles).

~/.hermes/
├── memories/                          # Persistent agent memories (markdown or JSON)
│   └── *.md / *.json
├── skills/                            # Installed skills (read-only for ingest purposes)
│   └── <skill-name>/SKILL.md
├── sessions/                          # Session transcripts (if session logging is enabled)
│   └── YYYY-MM-DD/
│       └── <session-id>.jsonl
├── config.yaml                        # User config (model, theme, paths)
└── .hub/                              # Skills Hub state (lock.json, audit.log, quarantine/)

Hermes将所有本地工件存储在

~/.hermes/

目录下（非默认配置文件则使用

$HERMES_HOME

）。

~/.hermes/
├── memories/                          # 持久化Agent记忆（Markdown或JSON格式）
│   └── *.md / *.json
├── skills/                            # 已安装的技能（导入时为只读）
│   └── <skill-name>/SKILL.md
├── sessions/                          # 会话记录（若启用会话日志）
│   └── YYYY-MM-DD/
│       └── <session-id>.jsonl
├── config.yaml                        # 用户配置（模型、主题、路径）
└── .hub/                              # 技能中心状态（lock.json、audit.log、quarantine/）

Key data sources ranked by value

按价值排序的关键数据源

```
memories/*.md
```
/
```
memories/*.json
```
— highest signal; curated persistent knowledge the agent accumulated
```
sessions/**/*.jsonl
```
— structured turn-by-turn transcripts; rich but noisy
```
config.yaml
```
— metadata only (model preferences, paths); rarely worth ingesting

Skip

.hub/

internals (audit/quarantine state) and the

skills/

directory (source material, not user knowledge).

```
memories/*.md
```
/
```
memories/*.json
```
—— 信号最强；Agent积累的经过整理的持久化知识
```
sessions/**/*.jsonl
```
—— 结构化的逐轮会话记录；信息丰富但噪音较多
```
config.yaml
```
—— 仅元数据（模型偏好、路径）；很少值得导入

跳过

.hub/

内部文件（审核/隔离状态）和

skills/

目录（源材料，非用户知识）。

Step 1: Survey and Compute Delta

步骤1：排查并计算增量

Scan

HERMES_HISTORY_PATH

and compare against

.manifest.json

```
~/.hermes/memories/
```
```
~/.hermes/sessions/**/
```
(if present)

Classify each file:

New — not in manifest
Modified — in manifest but file is newer than
```
ingested_at
```
Unchanged — already ingested and unchanged

Report a concise delta summary before deep parsing.

扫描

HERMES_HISTORY_PATH

并与

.manifest.json

对比：

```
~/.hermes/memories/
```
```
~/.hermes/sessions/**/
```
（若存在）

对每个文件进行分类：

新建 —— 未在清单中
已修改 —— 在清单中但文件比
```
ingested_at
```
更新
未更改 —— 已导入且未修改

在深度解析前，先报告简洁的增量摘要。

Step 2: Parse Memories First

步骤2：优先解析记忆

Memories are the highest-value source. Hermes writes them as either:

Markdown — structured prose with optional frontmatter; ingest directly

JSON —

{"content": "...", "created_at": "...", "tags": [...]}

records

For each memory:

Extract the core knowledge claim
Note any tags Hermes attached (they often map to wiki categories)
Merge into the appropriate wiki page rather than creating one memory = one page

记忆是最高价值的数据源。Hermes会以以下两种格式写入记忆：

Markdown —— 带可选前置元数据的结构化散文；直接导入

JSON ——

{"content": "...", "created_at": "...", "tags": [...]}

格式的记录

对于每条记忆：

提取核心知识主张
记录Hermes附加的标签（通常对应wiki分类）
合并到合适的wiki页面，而非一条记忆对应一个页面

Step 3: Parse Session JSONL Safely

步骤3：安全解析会话JSONL

Each session JSONL line is an event envelope. Common shapes:

json

{"role": "user", "content": "..."}
{"role": "assistant", "content": "..."}
{"type": "tool_use", "name": "...", "input": {...}}
{"type": "tool_result", "content": "..."}

每个会话JSONL的行都是一个事件包。常见格式如下：

json

{"role": "user", "content": "..."}
{"role": "assistant", "content": "..."}
{"type": "tool_use", "name": "...", "input": {...}}
{"type": "tool_result", "content": "..."}

Extraction rules

提取规则

Prioritize assistant responses that state conclusions, patterns, or decisions
Extract user intent from high-signal turns; skip low-information follow-ups
Treat
```
tool_use
```
/
```
tool_result
```
pairs as context, not primary content
Skip token accounting, internal plumbing, and repeated plan echoes

优先提取助手给出的结论、模式或决策相关的回复
从高信号交互中提取用户意图；跳过低信息量的后续跟进内容
将
```
tool_use
```
/
```
tool_result
```
对视为上下文，而非主要内容
跳过令牌统计、内部流程和重复的计划回声

Critical privacy filter

关键隐私过滤

Session logs can include injected instructions, tool payloads, and sensitive text. Do not ingest verbatim.

Remove API keys, tokens, passwords, credentials
Redact private identifiers unless relevant and user-approved
Summarize; do not quote raw transcripts verbatim

会话日志可能包含注入的指令、工具负载和敏感文本。请勿直接导入原文。

删除API密钥、令牌、密码、凭证
编辑私人标识符，除非相关且获得用户批准
进行总结；不要引用原始会话记录原文

Step 4: Cluster by Topic

步骤4：按主题聚类

Do not create one wiki page per memory or session.

Group memories by stable topic (concept, tool, project, technique)
Split mixed sessions into separate themes
Merge recurring patterns across dates and projects
Use file paths or session
```
cwd
```
metadata to infer project scope when available

不要为每条记忆或每个会话创建一个wiki页面。

按稳定主题（概念、工具、项目、技术）分组记忆
将混合主题的会话拆分为不同主题
合并跨日期和项目的重复模式
若有可用的文件路径或会话
```
cwd
```
元数据，用其推断项目范围

Step 5: Distill into Wiki Pages

步骤5：提炼为Wiki页面

Route extracted knowledge using existing wiki conventions:

Project-specific architecture/process →
```
projects/<name>/...
```
General concepts →
```
concepts/
```
Recurring techniques/debug playbooks →
```
skills/
```
Tools/services/frameworks →
```
entities/
```
Cross-session patterns →
```
synthesis/
```

For each impacted project, create/update

projects/<name>/<name>.md

使用现有wiki规范路由提取到的知识：

特定项目的架构/流程 →
```
projects/<name>/...
```
通用概念 →
```
concepts/
```
重复使用的技术/调试手册 →
```
skills/
```
工具/服务/框架 →
```
entities/
```
跨会话模式 →
```
synthesis/
```

对于每个受影响的项目，创建/更新

projects/<name>/<name>.md

。

Writing rules

写作规则

Distill knowledge, not chronology
Avoid "on date X we discussed..." unless date context is essential
Add
```
summary:
```
frontmatter on each new/updated page (1–2 sentences, ≤ 200 chars)
Add provenance markers:
- ```
^[extracted]
```
  when directly grounded in explicit memory/session content
- ```
^[inferred]
```
  when synthesizing patterns across multiple memories
- ```
^[ambiguous]
```
  when memories conflict
Add/update
```
provenance:
```
frontmatter mix for each changed page

提炼知识，而非记录时间线
避免使用“在X日期我们讨论了...”，除非日期上下文至关重要
在每个新建/更新的页面添加
```
summary:
```
前置元数据（1-2句话，≤200字符）
添加来源标记：
- ```
^[extracted]
```
  ：直接基于明确的记忆/会话内容
- ```
^[inferred]
```
  ：综合多条记忆的模式得出
- ```
^[ambiguous]
```
  ：记忆存在冲突时
为每个修改过的页面添加/更新
```
provenance:
```
前置元数据

Step 6: Update Manifest, Log, and Index

步骤6：更新清单、日志和索引

Update

.manifest.json

更新

.manifest.json

For each processed source file:

```
ingested_at
```
,
```
size_bytes
```
,
```
modified_at
```
```
source_type
```
:
```
hermes_memory
```
|
```
hermes_session
```
```
project
```
: inferred project name (when applicable)
```
pages_created
```
,
```
pages_updated
```

Add/update a top-level summary block:

json

{
  "hermes": {
    "source_path": "~/.hermes/",
    "last_ingested": "TIMESTAMP",
    "memories_ingested": 42,
    "sessions_ingested": 7,
    "pages_created": 5,
    "pages_updated": 12
  }
}

对于每个已处理的源文件：

```
ingested_at
```
、
```
size_bytes
```
、
```
modified_at
```
```
source_type
```
:
```
hermes_memory
```
|
```
hermes_session
```
```
project
```
: 推断的项目名称（若适用）
```
pages_created
```
、
```
pages_updated
```

添加/更新顶层摘要块：

json

{
  "hermes": {
    "source_path": "~/.hermes/",
    "last_ingested": "TIMESTAMP",
    "memories_ingested": 42,
    "sessions_ingested": 7,
    "pages_created": 5,
    "pages_updated": 12
  }
}

Update special files

更新特殊文件

Update

index.md

and

log.md

- [TIMESTAMP] HERMES_HISTORY_INGEST memories=N sessions=M pages_updated=X pages_created=Y mode=append|full

更新

index.md

和

log.md

：

- [TIMESTAMP] HERMES_HISTORY_INGEST memories=N sessions=M pages_updated=X pages_created=Y mode=append|full

Privacy and Compliance

隐私与合规

Distill and synthesize; avoid raw memory or transcript dumps
Default to redaction for anything that looks sensitive
Ask the user before storing personal or sensitive details
Keep references to other people minimal and purpose-bound

提炼和综合内容；避免直接导入原始记忆或会话记录
对任何看起来敏感的内容默认进行编辑
在存储个人或敏感信息前询问用户
尽量减少对他人的引用，且仅用于特定目的

Reference

参考

See

references/hermes-data-format.md

for field-level notes and extraction guidance.

查看

references/hermes-data-format.md

获取字段级说明和提取指南。

hermes-history-ingest

Original

Translation

Hermes History Ingest — Conversation & Memory Mining

Hermes历史记录导入——对话与记忆挖掘

Before You Start

开始之前

Ingest Modes

导入模式

Append Mode (default)

追加模式（默认）

Full Mode

全量模式

Hermes Data Layout

Hermes数据结构

Key data sources ranked by value

按价值排序的关键数据源

Step 1: Survey and Compute Delta

步骤1：排查并计算增量

Step 2: Parse Memories First

步骤2：优先解析记忆

Step 3: Parse Session JSONL Safely

步骤3：安全解析会话JSONL

Extraction rules

提取规则

Critical privacy filter

关键隐私过滤

Step 4: Cluster by Topic

步骤4：按主题聚类

Step 5: Distill into Wiki Pages

步骤5：提炼为Wiki页面

Writing rules

写作规则

Step 6: Update Manifest, Log, and Index

步骤6：更新清单、日志和索引

Update .manifest.json

更新.manifest.json

Update special files

更新特殊文件

Privacy and Compliance

隐私与合规

Reference

参考

Update
`.manifest.json`

更新
`.manifest.json`