中文 EN

Scan Report

Scan New Files

⚠ Low Risk — Risk Score 15/100
Last scan:23 hr ago Rescan
15 /100
pii-redactor
Redact sensitive information from text using a locally-hosted zero-shot PII/PHI detection model.
This skill is pure documentation with no implementation code. It describes a legitimate PII-redaction workflow with comprehensive security guidance, but relies on an unverifiable external package (clawguard-pii) with a placeholder GitHub source.
Skill Namepii-redactor
Duration36.2s
Enginepi
✓
Safe to install
Before installing clawguard-pii, audit the package source and verify the GitHub repository is legitimate. Add version-pinned checksums for the package binary.

Findings 2 items

Severity Finding Location
Medium
External package source unverifiable Supply Chain
The skill installs 'clawguard-pii==1.0.4' with a GitHub source URL that is explicitly marked as a placeholder. No PyPI listing, no real repository URL, no checksum. The package cannot be audited without verifiable source.
pip install clawguard-pii==1.0.4
# Source: https://github.com/anthropics/clawguard-pii (replace with the actual repository URL)
→ Replace placeholder URL with the actual audited repository. Verify package integrity via signed releases or pinned checksums before installation.
 SKILL.md:14
Low
Installation URL is a placeholder Doc Mismatch
The 'Source' field in the setup instructions is a placeholder that must be replaced. Users following the docs blindly would install from an unspecified source.
(replace with the actual repository URL)
→ Provide the real, verified repository URL before publishing this skill.
 SKILL.md:18
ResourceDeclaredInferredStatusEvidence
Filesystem NONE NONE — No implementation files present — only documentation.
Network READ READ ✓ Aligned SKILL.md declares POST to CLAWGUARD_URL/redact and GET /health — legitimately sc…
Shell NONE NONE — No shell commands executed in skill code.
Environment READ READ ✓ Aligned SKILL.md declares reading CLAWGUARD_URL and CLAWGUARD_TOKEN — necessary for serv…
Skill Invoke NONE NONE — No nested skill invocations found.
3 findings
🔗
Medium External URL 外部 URL
http://127.0.0.1:
SKILL.md:62
🔗
Medium External URL 外部 URL
https://developer.nvidia.com/open-model-license
SKILL.md:160
🔗
Medium External URL 外部 URL
https://spdx.org/licenses/MIT-0.html
SKILL.md:161

File Tree

1 files · 5.6 KB · 160 lines
Markdown 1f · 160L
└─ 📝 SKILL.md Markdown 160L · 5.6 KB

Dependencies 1 items

PackageVersionSourceKnown VulnsNotes
clawguard-pii 1.0.4 GitHub (placeholder URL — unverifiable) No GitHub URL is explicitly marked as placeholder; package source cannot be audited

Security Positives

✓ No implementation code present — skill is purely declarative documentation
✓ Comprehensive security documentation: URL validation rules, HTTPS requirements, token handling guidance
✓ No hidden functionality — all behavior is documented in SKILL.md
✓ No credential harvesting, no data exfiltration, no obfuscation, no base64 or shell execution
✓ Authorization token uses secrets.token_hex(32) — cryptographically appropriate
✓ Fallback manual-review path prevents silent data leaks when service is down
✓ The skill properly scopes network access to localhost/internal hosts only
✓ MIT-0 license clearly declared