security-check-skill

Name: security-check-skill
Author: berabuddies

555stars59forks

How to install security-check-skill?

berabuddies/semia/security-check-skill

$npx -y skills add berabuddies/Semia --skill security-check-skill

Installs into the current project.

Summary

Security audit and inspection skill for Clawdbot skills. Use this when you need to check skills for security vulnerabilities before installation, perform regular security audits on installed skills, verify skill description matches actual behavior, scan for prompt injection attem

Files · 1

View on GitHub

SKILL.md

1	# Security Check Skill
2
3	Comprehensive security auditing for Clawdbot skills to detect malicious intent, prompt injection, secrets exposure, and misaligned behavior.
4
5	## Quick Start
6
7	### Pre-Installation Security Check
8
9	Before installing a new skill from ClawdHub or any source:
10
11	1. Download and inspect the skill files
12	2. Run the automated security scanner:
13	```bash
14	python3 scripts/scan_skill.py /path/to/skill
15	```
16	3. Review the scanner output - Block any skill with HIGH severity issues
17	4. Manual review for MEDIUM severity issues
18	5. Verify behavior matches description before installation
19
20	### Daily Security Audit
21
22	Run daily to ensure installed skills remain secure:
23
24	```bash
25	# Scan all skills in the skills directory
26	python3 scripts/scan_skill.py /path/to/skills/skill-1
27	python3 scripts/scan_skill.py /path/to/skills/skill-2
28	# ... repeat for each installed skill
29	```
30
31	## Security Scanner
32
33	### Running the Scanner
34
35	The `scripts/scan_skill.py` tool provides automated security analysis:
36
37	```bash
38	python3 scripts/scan_skill.py <skill-path>
39	```
40
41	Output includes:
42	- HIGH severity issues (immediate action required)
43	- MEDIUM severity warnings (review recommended)
44	- LOW severity notes (informational)
45	- Summary of checks performed
46
47	Example output:
48	```json
49	{
50	"skill_name": "example-skill",
51	"issues": [
52	{
53	"severity": "HIGH",
54	"file": "SKILL.md",
55	"issue": "Potential prompt injection pattern",
56	"recommendation": "Review and remove suspicious patterns"
57	}
58	],
59	"warnings": [
60	{
61	"severity": "MEDIUM",
62	"file": "scripts/helper.py",
63	"issue": "os.system() usage detected",
64	"recommendation": "Review and ensure this is safe"
65	}
66	],
67	"passed": [
68	{"file": "SKILL.md", "check": "Prompt injection scan", "status": "Completed"}
69	],
70	"summary": "SECURITY ISSUES FOUND: 1 issue(s), 1 warning(s)"
71	}
72	```
73
74	### What the Scanner Checks
75
76	1. SKILL.md Analysis
77	- Prompt injection patterns
78	- External network calls
79	- Suspicious instructions
80
81	2. Scripts Directory Scan
82	- Dangerous command patterns (rm -rf, eval, exec)
83	- Hardcoded secrets and credentials
84	- Unsafe subprocess usage
85	- File system operations outside skill directory
86
87	3. References Directory Scan
88	- Hardcoded secrets (passwords, API keys, tokens)
89	- Suspicious URLs (pastebin, raw GitHub links)
90	- Sensitive information exposure
91
92	## Manual Security Checklist
93
94	Use the comprehensive checklist in `references/security-checklist.md` for manual reviews.
95
96	### Critical Checks (Before Installation)
97
98	#### 1. Documentation Integrity (SKILL.md)
99	- ✅ Description accurately reflects skill functionality
100	- ❌ No prompt injection patterns (see `references/prompt-injection-patterns.md`)
101	- ❌ No instructions to ignore/discard context
102	- ❌ No system override commands
103	- ✅ No hidden capabilities beyond description
104
105	#### 2. Code Review (scripts/)
106	- ❌ No hardcoded credentials or secrets
107	- ❌ No dangerous file operations (rm -rf outside skill dir)
108	- ❌ No eval() or exec() with user input
109	- ❌ No unauthorized network requests
110	- ✅ All operations within skill directory
111	- ✅ Proper input validation
112
113	#### 3. Reference Materials (references/)
114	- ❌ No hardcoded passwords, API keys, or tokens
115	- ❌ No production credentials in documentation
116	- ✅ Links only to legitimate, trusted sources
117	- ✅ No documentation of security bypasses
118
119	#### 4. Behavior Alignment
120	- ✅ Every command matches stated purpose
121	- ✅ No hidden capabilities
122	- ✅ No unnecessary file system access
123	- ✅ Network access only when explicitly required
124
125	### Daily Audit Checks
126
127	1. Scan all installed skills with the automated scanner
128	2. Review any new HIGH severity issues
129	3. Check for modified files in skill directories
130	4. Verify skill descriptions still match behavior
131	5. Audit new dependencies if added
132
133	## Specific Security Concerns
134
135	### Prompt Injection Detection
136
137	Read `references/prompt-injection-patterns.md` for comprehensive patterns.
138
139	Key indicators:
140	- Instructions to ignore/discard context
141	- System override or bypass commands
142	- Authority impersonation (act as administrator, etc.)
143	- Jailbreak attempts (unrestricted mode, etc.)
144	- Instruction replacement patterns
145
146	Detection:
147	```python
148	# Automated pattern matching
149	import r

People who liked this also viewed

6 picks

Type

security-check-skill

555stars59forks

View on GitHub

How to install security-check-skill?

berabuddies/semia/security-check-skill

$npx -y skills add berabuddies/Semia --skill security-check-skill

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Security Check Skill
2
3	Comprehensive security auditing for Clawdbot skills to detect malicious intent, prompt injection, secrets exposure, and misaligned behavior.
4
5	## Quick Start
6
7	### Pre-Installation Security Check
8
9	Before installing a new skill from ClawdHub or any source:
10
11	1. Download and inspect the skill files
12	2. Run the automated security scanner:
13	```bash
14	python3 scripts/scan_skill.py /path/to/skill
15	```
16	3. Review the scanner output - Block any skill with HIGH severity issues
17	4. Manual review for MEDIUM severity issues
18	5. Verify behavior matches description before installation
19
20	### Daily Security Audit
21
22	Run daily to ensure installed skills remain secure:
23
24	```bash
25	# Scan all skills in the skills directory
26	python3 scripts/scan_skill.py /path/to/skills/skill-1
27	python3 scripts/scan_skill.py /path/to/skills/skill-2
28	# ... repeat for each installed skill
29	```
30
31	## Security Scanner
32
33	### Running the Scanner
34
35	The `scripts/scan_skill.py` tool provides automated security analysis:
36
37	```bash
38	python3 scripts/scan_skill.py <skill-path>
39	```
40
41	Output includes:
42	- HIGH severity issues (immediate action required)
43	- MEDIUM severity warnings (review recommended)
44	- LOW severity notes (informational)
45	- Summary of checks performed
46
47	Example output:
48	```json
49	{
50	"skill_name": "example-skill",
51	"issues": [
52	{
53	"severity": "HIGH",
54	"file": "SKILL.md",
55	"issue": "Potential prompt injection pattern",
56	"recommendation": "Review and remove suspicious patterns"
57	}
58	],
59	"warnings": [
60	{
61	"severity": "MEDIUM",
62	"file": "scripts/helper.py",
63	"issue": "os.system() usage detected",
64	"recommendation": "Review and ensure this is safe"
65	}
66	],
67	"passed": [
68	{"file": "SKILL.md", "check": "Prompt injection scan", "status": "Completed"}
69	],
70	"summary": "SECURITY ISSUES FOUND: 1 issue(s), 1 warning(s)"
71	}
72	```
73
74	### What the Scanner Checks
75
76	1. SKILL.md Analysis
77	- Prompt injection patterns
78	- External network calls
79	- Suspicious instructions
80
81	2. Scripts Directory Scan
82	- Dangerous command patterns (rm -rf, eval, exec)
83	- Hardcoded secrets and credentials
84	- Unsafe subprocess usage
85	- File system operations outside skill directory
86
87	3. References Directory Scan
88	- Hardcoded secrets (passwords, API keys, tokens)
89	- Suspicious URLs (pastebin, raw GitHub links)
90	- Sensitive information exposure
91
92	## Manual Security Checklist
93
94	Use the comprehensive checklist in `references/security-checklist.md` for manual reviews.
95
96	### Critical Checks (Before Installation)
97
98	#### 1. Documentation Integrity (SKILL.md)
99	- ✅ Description accurately reflects skill functionality
100	- ❌ No prompt injection patterns (see `references/prompt-injection-patterns.md`)
101	- ❌ No instructions to ignore/discard context
102	- ❌ No system override commands
103	- ✅ No hidden capabilities beyond description
104
105	#### 2. Code Review (scripts/)
106	- ❌ No hardcoded credentials or secrets
107	- ❌ No dangerous file operations (rm -rf outside skill dir)
108	- ❌ No eval() or exec() with user input
109	- ❌ No unauthorized network requests
110	- ✅ All operations within skill directory
111	- ✅ Proper input validation
112
113	#### 3. Reference Materials (references/)
114	- ❌ No hardcoded passwords, API keys, or tokens
115	- ❌ No production credentials in documentation
116	- ✅ Links only to legitimate, trusted sources
117	- ✅ No documentation of security bypasses
118
119	#### 4. Behavior Alignment
120	- ✅ Every command matches stated purpose
121	- ✅ No hidden capabilities
122	- ✅ No unnecessary file system access
123	- ✅ Network access only when explicitly required
124
125	### Daily Audit Checks
126
127	1. Scan all installed skills with the automated scanner
128	2. Review any new HIGH severity issues
129	3. Check for modified files in skill directories
130	4. Verify skill descriptions still match behavior
131	5. Audit new dependencies if added
132
133	## Specific Security Concerns
134
135	### Prompt Injection Detection
136
137	Read `references/prompt-injection-patterns.md` for comprehensive patterns.
138
139	Key indicators:
140	- Instructions to ignore/discard context
141	- System override or bypass commands
142	- Authority impersonation (act as administrator, etc.)
143	- Jailbreak attempts (unrestricted mode, etc.)
144	- Instruction replacement patterns
145
146	Detection:
147	```python
148	# Automated pattern matching
149	import r

People who liked this also viewed

6 picks

Type