agent-init

Name: agent-init
Author: szsip239

101stars14forks

How to install agent-init?

szsip239/teamclaw/agent-init

$npx -y skills add szsip239/teamclaw --skill agent-init

Installs into the current project.

Summary

Initialize and configure OpenClaw agent workspace MD files (AGENTS.md, SOUL.md, IDENTITY.md, USER.md, TOOLS.md, BOOTSTRAP.md, HEARTBEAT.md). Use when: setting up a new agent, customizing agent personality/behavior, configuring agent workspace for TeamClaw, or checking/fixing agen

Files · 1

View on GitHub

SKILL.md

1	# Agent Init
2
3	Initialize OpenClaw agent workspace with tailored MD files through an interactive interview.
4
5	## Workflow
6
7	### Phase 1: Interview (MANDATORY — do not skip)
8
9	Before generating ANY files, gather context through conversation. Ask in batches of 2-3 questions:
10
11	Batch 1 — Identity & Purpose:
12	- What is this agent's primary purpose? (e.g., coding assistant, research, DevOps, personal assistant)
13	- What name and emoji? Any personality traits?
14
15	Batch 2 — User Profile:
16	- Who will use this agent? (name, timezone, preferences)
17	- Communication style preference? (formal/casual, verbose/terse, language)
18
19	Batch 3 — Environment:
20	- Is this a container instance or external instance?
21	- What tools/languages does the agent need? (Python, Node, etc.)
22	- Any specific workflows or periodic tasks?
23
24	Batch 4 — Boundaries:
25	- Any topics or actions the agent should avoid?
26	- Privacy requirements beyond defaults?
27
28	Only proceed to Phase 2 after user confirms the interview is complete.
29
30	### Phase 2: Environment Check
31
32	Run `scripts/check-env.sh` to detect Python/uv status:
33
34	```bash
35	bash <skill-path>/scripts/check-env.sh
36	```
37
38	If uv is missing and user wants Python support:
39	```bash
40	bash <skill-path>/scripts/check-env.sh --install
41	```
42
43	For container instances, run inside the container:
44	```bash
45	docker exec <containerId> bash -c "which uv && uv --version \|\| echo 'uv: NOT FOUND'"
46	```
47
48	### Phase 3: Generate Files
49
50	Generate files in this order, showing each to user for confirmation before writing:
51
52	1. IDENTITY.md — Fill in fields from interview
53	2. USER.md — Fill in user profile
54	3. SOUL.md — Rewrite content, keep 4-section structure (Core Truths / Boundaries / Vibe / Continuity)
55	4. TOOLS.md — Add environment info, Python/uv config, documentation access notes
56	5. AGENTS.md — Extend default template with domain-specific sections (see strategy below)
57	6. HEARTBEAT.md — Add periodic tasks if any
58	7. BOOTSTRAP.md — Skip unless user wants first-run ritual
59
60	#### AGENTS.md Strategy: Extend, Don't Replace
61
62	Read the current AGENTS.md first. The default template contains critical infrastructure:
63	- Session startup sequence (file loading order)
64	- Memory system (daily + MEMORY.md)
65	- Safety rules
66	- Heartbeat logic
67
68	Add new sections; never remove existing ones. Safe insertion points:
69	- After "Every Session" → domain-specific startup tasks
70	- After "Safety" → TeamClaw security rules + file exchange rules
71	- After "Tools" → Python/uv preferences, documentation access
72	- Before "Make It Yours" → project-specific workflows
73
74	#### Mandatory AGENTS.md Additions
75
76	Always add these to AGENTS.md regardless of agent type:
77
78	```markdown
79	## File Exchange (TeamClaw)
80	- User uploads appear in `current-session/input/`
81	- Save output files to `current-session/output/`
82	- Check input folder at session start for uploaded files
83	- ALWAYS use `current-session/` symlink — NEVER navigate sessions/ by folder name
84	(folders use TeamClaw DB IDs which differ from your OpenClaw session UUID)
85
86	## Security (TeamClaw)
87	- NEVER delete files you didn't create — use `trash` over `rm`
88	- NEVER display API keys, tokens, passwords, or credentials
89	- NEVER read other users' session directories
90	- NEVER use `web_fetch` tool (DNS issues with proxy) — use browser tools instead
91	- For OpenClaw docs: `qmd query "<topic>"` or `qmd get "openclaw-docs/<path>"`
92	```
93
94	#### Mandatory TOOLS.md Additions
95
96	```markdown
97	## Python
98	- Package manager: `uv` (NEVER use pip directly)
99	- Create venv: `uv venv .venv`
100	- Install: `uv pip install <package>`
101	- Run: `uv run python script.py`
102	- If uv missing: `curl -LsSf https://astral.sh/uv/install.sh \| sh`
103
104	## Documentation
105	- OpenClaw docs: `qmd query "<topic>"` or `qmd get "openclaw-docs/<path>"`
106	- Web browsing: use browser tools only (no web_fetch)
107	```
108
109	### Phase 4: Write Files
110
111	Determine write method based on instance type:
112
113	External instance (workspacePath set):
114	```bash
115	# Workspace files live at {workspacePath}/workspace/ (or workspace-{profile}/)
116	cat > {workspacePath}/workspace/SOUL.md << 'ENDOFFILE'
117	[content]
118	ENDOFFILE
119	```
120
121	Container instance (containerId set):
122	```bash
123	docker exec -i <containerId> sh -c 'cat > /home/node/.openclaw/workspace/SOUL.md' << 'ENDOFFILE'
124	[content]
125	ENDOFFILE
126	```
127
128	For non-main agents (agentId ≠ "main"):
129	- Check if agent has a dedicated workspace via `config.get`
130	- Agent workspace might be at `workspace-{agentId}/` or configured separately
131
132	### Phase 5: Verify
133
134	After writin

People who liked this also viewed

6 picks

Type

agent-init

101stars14forks

View on GitHub

How to install agent-init?

szsip239/teamclaw/agent-init

$npx -y skills add szsip239/teamclaw --skill agent-init

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Agent Init
2
3	Initialize OpenClaw agent workspace with tailored MD files through an interactive interview.
4
5	## Workflow
6
7	### Phase 1: Interview (MANDATORY — do not skip)
8
9	Before generating ANY files, gather context through conversation. Ask in batches of 2-3 questions:
10
11	Batch 1 — Identity & Purpose:
12	- What is this agent's primary purpose? (e.g., coding assistant, research, DevOps, personal assistant)
13	- What name and emoji? Any personality traits?
14
15	Batch 2 — User Profile:
16	- Who will use this agent? (name, timezone, preferences)
17	- Communication style preference? (formal/casual, verbose/terse, language)
18
19	Batch 3 — Environment:
20	- Is this a container instance or external instance?
21	- What tools/languages does the agent need? (Python, Node, etc.)
22	- Any specific workflows or periodic tasks?
23
24	Batch 4 — Boundaries:
25	- Any topics or actions the agent should avoid?
26	- Privacy requirements beyond defaults?
27
28	Only proceed to Phase 2 after user confirms the interview is complete.
29
30	### Phase 2: Environment Check
31
32	Run `scripts/check-env.sh` to detect Python/uv status:
33
34	```bash
35	bash <skill-path>/scripts/check-env.sh
36	```
37
38	If uv is missing and user wants Python support:
39	```bash
40	bash <skill-path>/scripts/check-env.sh --install
41	```
42
43	For container instances, run inside the container:
44	```bash
45	docker exec <containerId> bash -c "which uv && uv --version \|\| echo 'uv: NOT FOUND'"
46	```
47
48	### Phase 3: Generate Files
49
50	Generate files in this order, showing each to user for confirmation before writing:
51
52	1. IDENTITY.md — Fill in fields from interview
53	2. USER.md — Fill in user profile
54	3. SOUL.md — Rewrite content, keep 4-section structure (Core Truths / Boundaries / Vibe / Continuity)
55	4. TOOLS.md — Add environment info, Python/uv config, documentation access notes
56	5. AGENTS.md — Extend default template with domain-specific sections (see strategy below)
57	6. HEARTBEAT.md — Add periodic tasks if any
58	7. BOOTSTRAP.md — Skip unless user wants first-run ritual
59
60	#### AGENTS.md Strategy: Extend, Don't Replace
61
62	Read the current AGENTS.md first. The default template contains critical infrastructure:
63	- Session startup sequence (file loading order)
64	- Memory system (daily + MEMORY.md)
65	- Safety rules
66	- Heartbeat logic
67
68	Add new sections; never remove existing ones. Safe insertion points:
69	- After "Every Session" → domain-specific startup tasks
70	- After "Safety" → TeamClaw security rules + file exchange rules
71	- After "Tools" → Python/uv preferences, documentation access
72	- Before "Make It Yours" → project-specific workflows
73
74	#### Mandatory AGENTS.md Additions
75
76	Always add these to AGENTS.md regardless of agent type:
77
78	```markdown
79	## File Exchange (TeamClaw)
80	- User uploads appear in `current-session/input/`
81	- Save output files to `current-session/output/`
82	- Check input folder at session start for uploaded files
83	- ALWAYS use `current-session/` symlink — NEVER navigate sessions/ by folder name
84	(folders use TeamClaw DB IDs which differ from your OpenClaw session UUID)
85
86	## Security (TeamClaw)
87	- NEVER delete files you didn't create — use `trash` over `rm`
88	- NEVER display API keys, tokens, passwords, or credentials
89	- NEVER read other users' session directories
90	- NEVER use `web_fetch` tool (DNS issues with proxy) — use browser tools instead
91	- For OpenClaw docs: `qmd query "<topic>"` or `qmd get "openclaw-docs/<path>"`
92	```
93
94	#### Mandatory TOOLS.md Additions
95
96	```markdown
97	## Python
98	- Package manager: `uv` (NEVER use pip directly)
99	- Create venv: `uv venv .venv`
100	- Install: `uv pip install <package>`
101	- Run: `uv run python script.py`
102	- If uv missing: `curl -LsSf https://astral.sh/uv/install.sh \| sh`
103
104	## Documentation
105	- OpenClaw docs: `qmd query "<topic>"` or `qmd get "openclaw-docs/<path>"`
106	- Web browsing: use browser tools only (no web_fetch)
107	```
108
109	### Phase 4: Write Files
110
111	Determine write method based on instance type:
112
113	External instance (workspacePath set):
114	```bash
115	# Workspace files live at {workspacePath}/workspace/ (or workspace-{profile}/)
116	cat > {workspacePath}/workspace/SOUL.md << 'ENDOFFILE'
117	[content]
118	ENDOFFILE
119	```
120
121	Container instance (containerId set):
122	```bash
123	docker exec -i <containerId> sh -c 'cat > /home/node/.openclaw/workspace/SOUL.md' << 'ENDOFFILE'
124	[content]
125	ENDOFFILE
126	```
127
128	For non-main agents (agentId ≠ "main"):
129	- Check if agent has a dedicated workspace via `config.get`
130	- Agent workspace might be at `workspace-{agentId}/` or configured separately
131
132	### Phase 5: Verify
133
134	After writin

People who liked this also viewed

6 picks

Type