中文 EN

Scan Report

Scan New Files

⚠ Low Risk — Risk Score 15/100
Last scan:1 day ago Rescan
15 /100
humanizerai
AI text detection and humanization CLI to bypass GPTZero, Turnitin, and other detectors
A legitimate third-party CLI wrapper for humanizerai.com API with no malicious behavior, though Bash tool wildcard access is broader than strictly necessary.
Skill Namehumanizerai
Duration41.7s
Enginepi
✓
Safe to install
Consider restricting Bash to specific subcommands (e.g., Bash(humanizerai:detect|humanize|credits)) to align with least-privilege principle.

Findings 2 items

Severity Finding Location
Low
Filesystem read not documented Doc Mismatch
SKILL.md does not mention the -f flag for file-based operations, though it's implemented and used legitimately in detect and humanize commands.
text = fs.readFileSync(argv.file, 'utf-8');
→ Document the -f flag in SKILL.md usage examples for completeness.
 src/commands/detect.ts:14
Low
Bash wildcard broader than necessary Priv Escalation
allowed-tools declares Bash(humanizerai:*) which technically allows any subcommand, though only detect/humanize/credits exist in the codebase.
allowed-tools: Bash(humanizerai:*)
→ Consider restricting to specific commands: Bash(humanizerai:detect|humanize|credits) for least-privilege alignment.
 SKILL.md:7
ResourceDeclaredInferredStatusEvidence
Network READ READ ✓ Aligned api.ts:35 - Makes HTTP POST/GET requests to humanizerai.com API
Filesystem NONE READ ✓ Aligned commands/detect.ts:14, commands/humanize.ts:14 - Uses fs.readFileSync with -f fl…
Environment READ READ ✓ Aligned config.ts:9 - Reads HUMANIZERAI_API_KEY and HUMANIZERAI_API_URL from process.env
Shell WRITE WRITE ✓ Aligned SKILL.md declares Bash(humanizerai:*) which provides shell:WRITE
12 findings
🔗
Medium External URL 外部 URL
https://humanizerai.com/dashboard
README.md:19
🔗
Medium External URL 外部 URL
https://humanizerai.com
README.md:72
🔗
Medium External URL 外部 URL
https://humanizerai.com/docs/api
README.md:73
🔗
Medium External URL 外部 URL
https://agentskill.sh
README.md:74
🔗
Medium External URL 外部 URL
https://www.npmjs.com/package/humanizerai
SKILL.md:34
🔗
Medium External URL 外部 URL
https://agentskill.sh**
SKILL.md:369
🔗
Medium External URL 外部 URL
https://humanizerai.com/api/v1
dist/index.js:34
🔗
Medium External URL 外部 URL
https://humanizerai.com\n\nSet
dist/index.js:212
🔗
Medium External URL 外部 URL
https://paulmillr.com/funding/
package-lock.json:971
🔗
Medium External URL 外部 URL
https://opencollective.com/postcss/
package-lock.json:1318
🔗
Medium External URL 外部 URL
https://humanizerai.com\n\n
src/index.ts:79
📧
Info Email 邮箱地址
[email protected]
package.json:35

File Tree

13 files · 80.8 KB · 2709 lines
JSON 3f · 1756L Markdown 2f · 443L TypeScript 7f · 297L JavaScript 1f · 213L
├─ ▾ 📁 dist
│ └─ 📜 index.js JavaScript 213L · 7.0 KB
├─ ▾ 📁 src
│ ├─ ▾ 📁 commands
│ │ ├─ 📜 credits.ts TypeScript 13L · 323 B
│ │ ├─ 📜 detect.ts TypeScript 34L · 916 B
│ │ └─ 📜 humanize.ts TypeScript 42L · 1.1 KB
│ ├─ 📜 api.ts TypeScript 96L · 2.2 KB
│ ├─ 📜 config.ts TypeScript 17L · 564 B
│ └─ 📜 index.ts TypeScript 83L · 2.6 KB
├─ 📋 package-lock.json JSON 1685L · 53.8 KB
├─ 📋 package.json JSON 56L · 1.2 KB
├─ 📝 README.md Markdown 74L · 1.7 KB
├─ 📝 SKILL.md Markdown 369L · 9.0 KB
├─ 📋 tsconfig.json JSON 15L · 304 B
└─ 📜 tsup.config.ts TypeScript 12L · 214 B

Dependencies 1 items

PackageVersionSourceKnown VulnsNotes
yargs ^17.7.2 npm No Popular CLI parsing library with stable versioning

Security Positives

✓ No credential exfiltration or API key harvesting beyond intended usage
✓ No base64-encoded payloads or obfuscated code
✓ No remote script execution (curl|bash, wget|sh)
✓ No access to sensitive paths (~/.ssh, ~/.aws, .env)
✓ No data exfiltration to external IPs beyond the documented API
✓ Clean, straightforward TypeScript implementation with no hidden functionality
✓ Dependencies are well-known packages (yargs) with reasonable version constraints
✓ API key is used only for authenticated requests to the documented humanizerai.com endpoint