hermes-memory-personality

Name: hermes-memory-personality
Author: timfewi

29stars2forks

How to install hermes-memory-personality?

timfewi/tentaflake/hermes-memory-personality

$npx -y skills add timfewi/tentaflake --skill hermes-memory-personality

Installs into the current project.

Summary

Manage Hermes memory (MEMORY.md, USER.md) and personality (SOUL.md, /personality) — persistence, capacity, context files, identity

Files · 1

View on GitHub

SKILL.md

1	# Hermes Memory & Personality
2
3	> Tentaflake context: This skill describes Hermes' built-in memory system
4	> — it works the same inside tentaflake agent containers. The memory settings
5	> (`memory.memory_enabled`, `memory.user_profile_enabled`, etc.) can be
6	> configured declaratively via the `settings` attrset in `my-agents.nix`.
7	> See `my-agents.nix.example` for examples. Agent state (including memory)
8	> persists in `/var/lib/hermes-<name>/` on the host.
9
10	## When to Use
11
12	- Manage what the agent remembers across sessions
13	- Configure SOUL.md personality
14	- Set up AGENTS.md project context files
15	- Control memory capacity and consolidation
16	- Switch personalities mid-session
17	- Understand memory vs skills distinction
18	- Use context files (CLAUDE.md, .cursorrules)
19
20	## Procedure
21
22	### 1. Memory System Overview
23
24	Two files persist across sessions:
25
26	\| File \| Purpose \| Char Limit \| Location \|
27	\| ------------- \| ----------------------------------------------- \| ------------------------- \| ------------------------------ \|
28	\| MEMORY.md \| Agent's notes — env facts, conventions, lessons \| 2,200 chars (~800 tokens) \| `~/.hermes/memories/MEMORY.md` \|
29	\| USER.md \| User profile — preferences, style, expectations \| 1,375 chars (~500 tokens) \| `~/.hermes/memories/USER.md` \|
30
31	Both inject into system prompt as frozen snapshot at session start.
32
33	### 2. Memory Tool Actions
34
35	Agent uses `memory` tool:
36
37	```text
38	memory(action="add", target="memory", content="User prefers TypeScript over JavaScript")
39	memory(action="replace", target="memory",
40	old_text="TypeScript",
41	content="User prefers TypeScript over JavaScript")
42	memory(action="remove", target="memory", old_text="TypeScript")
43	```
44
45	\| Action \| Use \| Key params \|
46	\| --------- \| --------------- \| ------------------------------------------- \|
47	\| `add` \| New entry \| `target` (`memory` / `user`), `content` \|
48	\| `replace` \| Update existing \| `target`, `old_text` (substring), `content` \|
49	\| `remove` \| Delete entry \| `target`, `old_text` (substring) \|
50
51	No `read` — memory auto-injected in system prompt. `old_text` uses substring matching. If multiple matches → error asking for more specific text.
52
53	### 3. Memory vs Skills
54
55	\| Feature \| Memory \| Skills \|
56	\| ---------- \| ------------------------------- \| -------------------------------- \|
57	\| Stores \| Facts ("what") \| Procedures ("how") \|
58	\| Trigger \| Auto-injected in prompt \| On-demand via `/skill-name` \|
59	\| Capacity \| ~1,300 tokens total \| Unlimited (files on disk) \|
60	\| Management \| Agent curated via `memory` tool \| Agent created via `skill_manage` \|
61	\| Token cost \| Fixed per session \| Variable (loaded when needed) \|
62
63	Memory = facts (environment, preferences). Skills = procedures (multi-step workflows).
64
65	### 4. Capacity Management
66
67	Char limits:
68
69	\| Store \| Limit \| Typical entries \|
70	\| ------ \| ----------- \| --------------- \|
71	\| memory \| 2,200 chars \| 8-15 entries \|
72	\| user \| 1,375 chars \| 5-10 entries \|
73
74	When memory is full, add returns error with current entries. Agent should:
75
76	1. Read current entries
77	2. Remove or consolidate
78	3. Replace merged entries (shorter)
79	4. Re-add new entry
80
81	Best practice: Above 80% capacity, consolidate before adding. Merge related facts.
82
83	Good entries:
84
85	```text
86	# Packs multiple facts
87	User runs macOS 14 Sonoma, uses Homebrew, has Docker Desktop. Shell: zsh with oh-my-zsh.
88
89	# Specific convention
90	Project ~/code/api uses Go 1.22, sqlc for DB queries, chi router. Run tests with 'make test'.
91	```
92
93	Bad entries:
94
95	```text
96	# Too vague
97	User has a project.
98
99	# Too verbose
100	On January 5th, 2026, the user asked me to look at...
101	```
102
103	### 5. Memory Config
104
105	```yaml
106	# ~/.hermes/config.yaml
107	memory:
108	memory_enabled: true
109	user_profile_enabled: true
110	memory_char_limit: 2200
111	user_char_limit: 1375
112	write_approval: false # false=write freely \| true=require approval
113	```
114
115	`write_approval: true` gates all saves. Use `/memory` commands in session:
116
117	```text
118	/memory pending # List staged writes
119	/memory approve <id> # Apply one (or 'all')
120	/memory reject <id> # Drop one (or 'all')
121	/memory approval on # Turn gate on/off
122	```
123
124	### 6. Session Search vs Memory
125
126	\| Feature \| Memory \| Session Search \|
127	\| -------- \| -------------------------- \| ------------------- \|
128	\| Capacity \| ~1,30

People who liked this also viewed

6 picks

Type

lark-drive飞书云空间（云盘/云存储）：管理 Drive 文件和文件夹，包含上传/下载、创建文件夹、复制/移动/删除、查看元数据、评论/权限/订阅、标题、版本、飞书文档密级标签（secure labels）和本地文件导入。用户需要整理云盘目录、处理云空间资源 URL/token、判断链接类型/真实 token/标题，或导入 Word/Markdown/Excel/CSV/PPTX/.base 为 docx/sSkillsJul 2026382k16k
lark-sharedUse for lark-cli setup/auth tasks: auth login/status/logout, user vs bot identity, business-domain permissions (--domain, including all/docs/drive), missing scopes, revoking authorization, or handlingSkillsJul 2026382k16k
lark-whiteboard飞书画板：查询和编辑飞书云文档中的画板。支持导出画板为预览图片、导出原始节点结构、使用多种格式更新画板内容。当用户需要查看画板内容、导出画板图片、编辑画板时使用此 skill。不负责：飞书云文档内容编辑（lark-doc）、文档内嵌电子表格/Base（lark-sheets / lark-base）。SkillsJul 2026380k16k
lark-task飞书任务：管理任务、清单和任务智能体。创建待办任务、查看和更新任务状态、拆分子任务、组织任务清单、分配协作成员、上传任务附件、注册或注销任务智能体、更新任务智能体的主页数据、写入智能体任务记录。当用户需要创建待办事项、查看任务列表、跟踪任务进度、管理项目清单或给他人分配任务、为任务上传附件文件、注册注销任务智能体、更新智能体主页数据、写入任务记录时使用。SkillsJul 2026380k16k
lark-attendance飞书考勤打卡：查询自己的考勤打卡记录SkillsJul 2026330k16k
lark-apps妙搭（Spark/Miaoda）应用开发与托管：应用创建、本地全栈开发、云端生成迭代、创意设计（UI mockup / 可交互原型 / 线框图 / 落地页 / 仪表盘 / 幻灯片 deck / 视觉探索）、AI相关能力和飞书平台能力或者其他外部能力集成、日志/Trace/监控指标/PV/UV 查询、环境变量管理、应用角色与成员管理、自动化触发器（定时/记录变更/Webhook/飞书审批）。当用户要SkillsJul 2026227k16k

hermes-memory-personality

29stars2forks

View on GitHub

How to install hermes-memory-personality?

timfewi/tentaflake/hermes-memory-personality

$npx -y skills add timfewi/tentaflake --skill hermes-memory-personality

Installs into the current project.

Summary

Manage Hermes memory (MEMORY.md, USER.md) and personality (SOUL.md, /personality) — persistence, capacity, context files, identity

Files · 1

View on GitHub

SKILL.md

1	# Hermes Memory & Personality
2
3	> Tentaflake context: This skill describes Hermes' built-in memory system
4	> — it works the same inside tentaflake agent containers. The memory settings
5	> (`memory.memory_enabled`, `memory.user_profile_enabled`, etc.) can be
6	> configured declaratively via the `settings` attrset in `my-agents.nix`.
7	> See `my-agents.nix.example` for examples. Agent state (including memory)
8	> persists in `/var/lib/hermes-<name>/` on the host.
9
10	## When to Use
11
12	- Manage what the agent remembers across sessions
13	- Configure SOUL.md personality
14	- Set up AGENTS.md project context files
15	- Control memory capacity and consolidation
16	- Switch personalities mid-session
17	- Understand memory vs skills distinction
18	- Use context files (CLAUDE.md, .cursorrules)
19
20	## Procedure
21
22	### 1. Memory System Overview
23
24	Two files persist across sessions:
25
26	\| File \| Purpose \| Char Limit \| Location \|
27	\| ------------- \| ----------------------------------------------- \| ------------------------- \| ------------------------------ \|
28	\| MEMORY.md \| Agent's notes — env facts, conventions, lessons \| 2,200 chars (~800 tokens) \| `~/.hermes/memories/MEMORY.md` \|
29	\| USER.md \| User profile — preferences, style, expectations \| 1,375 chars (~500 tokens) \| `~/.hermes/memories/USER.md` \|
30
31	Both inject into system prompt as frozen snapshot at session start.
32
33	### 2. Memory Tool Actions
34
35	Agent uses `memory` tool:
36
37	```text
38	memory(action="add", target="memory", content="User prefers TypeScript over JavaScript")
39	memory(action="replace", target="memory",
40	old_text="TypeScript",
41	content="User prefers TypeScript over JavaScript")
42	memory(action="remove", target="memory", old_text="TypeScript")
43	```
44
45	\| Action \| Use \| Key params \|
46	\| --------- \| --------------- \| ------------------------------------------- \|
47	\| `add` \| New entry \| `target` (`memory` / `user`), `content` \|
48	\| `replace` \| Update existing \| `target`, `old_text` (substring), `content` \|
49	\| `remove` \| Delete entry \| `target`, `old_text` (substring) \|
50
51	No `read` — memory auto-injected in system prompt. `old_text` uses substring matching. If multiple matches → error asking for more specific text.
52
53	### 3. Memory vs Skills
54
55	\| Feature \| Memory \| Skills \|
56	\| ---------- \| ------------------------------- \| -------------------------------- \|
57	\| Stores \| Facts ("what") \| Procedures ("how") \|
58	\| Trigger \| Auto-injected in prompt \| On-demand via `/skill-name` \|
59	\| Capacity \| ~1,300 tokens total \| Unlimited (files on disk) \|
60	\| Management \| Agent curated via `memory` tool \| Agent created via `skill_manage` \|
61	\| Token cost \| Fixed per session \| Variable (loaded when needed) \|
62
63	Memory = facts (environment, preferences). Skills = procedures (multi-step workflows).
64
65	### 4. Capacity Management
66
67	Char limits:
68
69	\| Store \| Limit \| Typical entries \|
70	\| ------ \| ----------- \| --------------- \|
71	\| memory \| 2,200 chars \| 8-15 entries \|
72	\| user \| 1,375 chars \| 5-10 entries \|
73
74	When memory is full, add returns error with current entries. Agent should:
75
76	1. Read current entries
77	2. Remove or consolidate
78	3. Replace merged entries (shorter)
79	4. Re-add new entry
80
81	Best practice: Above 80% capacity, consolidate before adding. Merge related facts.
82
83	Good entries:
84
85	```text
86	# Packs multiple facts
87	User runs macOS 14 Sonoma, uses Homebrew, has Docker Desktop. Shell: zsh with oh-my-zsh.
88
89	# Specific convention
90	Project ~/code/api uses Go 1.22, sqlc for DB queries, chi router. Run tests with 'make test'.
91	```
92
93	Bad entries:
94
95	```text
96	# Too vague
97	User has a project.
98
99	# Too verbose
100	On January 5th, 2026, the user asked me to look at...
101	```
102
103	### 5. Memory Config
104
105	```yaml
106	# ~/.hermes/config.yaml
107	memory:
108	memory_enabled: true
109	user_profile_enabled: true
110	memory_char_limit: 2200
111	user_char_limit: 1375
112	write_approval: false # false=write freely \| true=require approval
113	```
114
115	`write_approval: true` gates all saves. Use `/memory` commands in session:
116
117	```text
118	/memory pending # List staged writes
119	/memory approve <id> # Apply one (or 'all')
120	/memory reject <id> # Drop one (or 'all')
121	/memory approval on # Turn gate on/off
122	```
123
124	### 6. Session Search vs Memory
125
126	\| Feature \| Memory \| Session Search \|
127	\| -------- \| -------------------------- \| ------------------- \|
128	\| Capacity \| ~1,30

People who liked this also viewed

6 picks

Type