deadclaw

Name: deadclaw
Author: berabuddies

555stars59forks

How to install deadclaw?

berabuddies/semia/deadclaw

$npx -y skills add berabuddies/Semia --skill deadclaw

Installs into the current project.

Summary

Emergency kill switch for OpenClaw agents. Instantly halts all running agents, pauses scheduled jobs, kills active sessions, and logs everything — triggered by a single message, a WebChat button, or a phone home screen shortcut. Includes a background watchdog that auto-kills agen

Files · 1

View on GitHub

SKILL.md

1	# DeadClaw — Emergency Kill Switch for OpenClaw Agents
2
3	> "One tap. Everything stops."
4
5	DeadClaw is a single-purpose emergency kill switch. When triggered, it immediately
6	halts all running OpenClaw agent processes, stops Docker containers, pauses all
7	scheduled cron jobs and heartbeats, kills all active sessions, and writes a
8	timestamped incident log. It then confirms back to whichever surface triggered it.
9
10	Works with both native OpenClaw installs and Docker-based deployments
11	(Hostinger VPS, etc.). Auto-detects your setup.
12
13	## Why DeadClaw Exists
14
15	The ClawHavoc attack (February 2026) exposed 1,184 malicious skills in the OpenClaw
16	ecosystem. People run agents autonomously — often overnight, often unattended. When
17	something goes wrong, you need a way to stop everything from wherever you are, even
18	from your phone. DeadClaw is that way.
19
20	---
21
22	## Activation Methods
23
24	DeadClaw supports three activation methods. All three execute the exact same kill
25	sequence — the only difference is how the user triggers it.
26
27	### Method 1: Message Trigger
28
29	The user sends a trigger word to any connected OpenClaw channel (Telegram, WhatsApp,
30	Discord, Slack, or any other connected channel). The following words activate DeadClaw:
31
32	- `kill` or `KILL`
33	- `dead`
34	- `stop everything`
35	- `emergency stop`
36	- `deadclaw`
37	- `🔴`
38
39	When a trigger word is detected:
40
41	1. Execute `scripts/kill.sh` from the DeadClaw skill directory
42	2. Capture the output (process count, containers stopped, cron jobs paused, timestamp)
43	3. Send confirmation back to the same channel the trigger came from:
44	`🔴 DeadClaw activated. All agents stopped. [timestamp] — [X] killed ([X] processes, [X] containers). [X] cron jobs paused. See deadclaw.log for full report.`
45
46	### Method 2: WebChat Kill Button
47
48	A persistent red button rendered in the OpenClaw WebChat dashboard. The HTML widget
49	is located at `ui/deadclaw-button.html`. It calls `kill.sh` via OpenClaw's WebChat
50	API hooks (`window.OpenClaw.exec()`). If the WebChat hooks are unavailable, the
51	button degrades to showing an error message with manual instructions.
52
53	To embed the button, use OpenClaw's WebChat customization hooks:
54
55	```javascript
56	OpenClaw.WebChat.registerWidget('deadclaw-button', {
57	src: 'skills/deadclaw/ui/deadclaw-button.html',
58	position: 'top-bar',
59	persistent: true
60	});
61	```
62
63	### Method 3: Phone Home Screen Shortcut
64
65	A pre-built shortcut that sends the kill trigger message (`deadclaw`) to the user's
66	configured Telegram bot. Setup guides for iOS and Android are in `docs/`:
67
68	- `docs/iphone-shortcut-guide.md` — iOS Shortcuts setup
69	- `docs/android-widget-guide.md` — Android widget setup (Tasker or HTTP Shortcuts)
70
71	---
72
73	## Watchdog (Passive Protection)
74
75	DeadClaw includes a background watchdog (`scripts/watchdog.sh`) that monitors for
76	dangerous conditions and auto-triggers the kill without any user action.
77
78	The watchdog checks every 60 seconds for (after a 5-minute startup grace period):
79
80	1. Runaway loops — Any agent process or Docker session running longer than 30 minutes
81	2. Token burn — Token spend exceeding 50,000 tokens in under 10 minutes
82	3. Unauthorized network — Outbound network calls to domains not on the user's whitelist
83	4. Sandbox escape — Any process attempting to write outside the designated OpenClaw workspace
84
85	The watchdog uses zero AI tokens — all checks use local system commands only.
86
87	When the watchdog auto-triggers, it sends an alert explaining the reason:
88	`🔴 DeadClaw auto-triggered. Reason: [specific reason]. All processes stopped. Check deadclaw.log.`
89
90	### Watchdog Configuration
91
92	The watchdog reads its thresholds from environment variables (with sensible defaults):
93
94	\| Variable \| Default \| Description \|
95	\|---\|---\|---\|
96	\| `DEADCLAW_MAX_RUNTIME_MIN` \| 30 \| Max agent runtime in minutes before auto-kill \|
97	\| `DEADCLAW_MAX_TOKENS` \| 50000 \| Max token spend in the monitoring window \|
98	\| `DEADCLAW_TOKEN_WINDOW_MIN` \| 10 \| Token spend monitoring window in minutes \|
99	\| `DEADCLAW_WHITELIST` \| `./network-whitelist.txt` \| Allowed outbound domains (one per line) \|
100	\| `DEADCLAW_WORKSPACE` \| `$OPENCLAW_WORKSPACE` \|

People who liked this also viewed

6 picks

Type

deadclaw

555stars59forks

View on GitHub

How to install deadclaw?

berabuddies/semia/deadclaw

$npx -y skills add berabuddies/Semia --skill deadclaw

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# DeadClaw — Emergency Kill Switch for OpenClaw Agents
2
3	> "One tap. Everything stops."
4
5	DeadClaw is a single-purpose emergency kill switch. When triggered, it immediately
6	halts all running OpenClaw agent processes, stops Docker containers, pauses all
7	scheduled cron jobs and heartbeats, kills all active sessions, and writes a
8	timestamped incident log. It then confirms back to whichever surface triggered it.
9
10	Works with both native OpenClaw installs and Docker-based deployments
11	(Hostinger VPS, etc.). Auto-detects your setup.
12
13	## Why DeadClaw Exists
14
15	The ClawHavoc attack (February 2026) exposed 1,184 malicious skills in the OpenClaw
16	ecosystem. People run agents autonomously — often overnight, often unattended. When
17	something goes wrong, you need a way to stop everything from wherever you are, even
18	from your phone. DeadClaw is that way.
19
20	---
21
22	## Activation Methods
23
24	DeadClaw supports three activation methods. All three execute the exact same kill
25	sequence — the only difference is how the user triggers it.
26
27	### Method 1: Message Trigger
28
29	The user sends a trigger word to any connected OpenClaw channel (Telegram, WhatsApp,
30	Discord, Slack, or any other connected channel). The following words activate DeadClaw:
31
32	- `kill` or `KILL`
33	- `dead`
34	- `stop everything`
35	- `emergency stop`
36	- `deadclaw`
37	- `🔴`
38
39	When a trigger word is detected:
40
41	1. Execute `scripts/kill.sh` from the DeadClaw skill directory
42	2. Capture the output (process count, containers stopped, cron jobs paused, timestamp)
43	3. Send confirmation back to the same channel the trigger came from:
44	`🔴 DeadClaw activated. All agents stopped. [timestamp] — [X] killed ([X] processes, [X] containers). [X] cron jobs paused. See deadclaw.log for full report.`
45
46	### Method 2: WebChat Kill Button
47
48	A persistent red button rendered in the OpenClaw WebChat dashboard. The HTML widget
49	is located at `ui/deadclaw-button.html`. It calls `kill.sh` via OpenClaw's WebChat
50	API hooks (`window.OpenClaw.exec()`). If the WebChat hooks are unavailable, the
51	button degrades to showing an error message with manual instructions.
52
53	To embed the button, use OpenClaw's WebChat customization hooks:
54
55	```javascript
56	OpenClaw.WebChat.registerWidget('deadclaw-button', {
57	src: 'skills/deadclaw/ui/deadclaw-button.html',
58	position: 'top-bar',
59	persistent: true
60	});
61	```
62
63	### Method 3: Phone Home Screen Shortcut
64
65	A pre-built shortcut that sends the kill trigger message (`deadclaw`) to the user's
66	configured Telegram bot. Setup guides for iOS and Android are in `docs/`:
67
68	- `docs/iphone-shortcut-guide.md` — iOS Shortcuts setup
69	- `docs/android-widget-guide.md` — Android widget setup (Tasker or HTTP Shortcuts)
70
71	---
72
73	## Watchdog (Passive Protection)
74
75	DeadClaw includes a background watchdog (`scripts/watchdog.sh`) that monitors for
76	dangerous conditions and auto-triggers the kill without any user action.
77
78	The watchdog checks every 60 seconds for (after a 5-minute startup grace period):
79
80	1. Runaway loops — Any agent process or Docker session running longer than 30 minutes
81	2. Token burn — Token spend exceeding 50,000 tokens in under 10 minutes
82	3. Unauthorized network — Outbound network calls to domains not on the user's whitelist
83	4. Sandbox escape — Any process attempting to write outside the designated OpenClaw workspace
84
85	The watchdog uses zero AI tokens — all checks use local system commands only.
86
87	When the watchdog auto-triggers, it sends an alert explaining the reason:
88	`🔴 DeadClaw auto-triggered. Reason: [specific reason]. All processes stopped. Check deadclaw.log.`
89
90	### Watchdog Configuration
91
92	The watchdog reads its thresholds from environment variables (with sensible defaults):
93
94	\| Variable \| Default \| Description \|
95	\|---\|---\|---\|
96	\| `DEADCLAW_MAX_RUNTIME_MIN` \| 30 \| Max agent runtime in minutes before auto-kill \|
97	\| `DEADCLAW_MAX_TOKENS` \| 50000 \| Max token spend in the monitoring window \|
98	\| `DEADCLAW_TOKEN_WINDOW_MIN` \| 10 \| Token spend monitoring window in minutes \|
99	\| `DEADCLAW_WHITELIST` \| `./network-whitelist.txt` \| Allowed outbound domains (one per line) \|
100	\| `DEADCLAW_WORKSPACE` \| `$OPENCLAW_WORKSPACE` \|

People who liked this also viewed

6 picks

Type