Securitysuite

• Outcome: Pass
• Latest audit: May 28, 2026 · 6:03 AM UTC
• Version: 1.0.0

Overview

This appears to be a local skill-auditing helper that sends user-selected skill files to a localhost server, with disclosure gaps but no artifact-backed malicious behavior.

Install only if you are comfortable running a local audit server and sending whole skill files to it. Keep it bound to localhost, avoid submitting files that contain secrets or proprietary material, and check whether the service logs request bodies before using batch mode.

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 92% confidence
• Finding: The skill declares no permissions, yet it clearly exposes and documents network-accessible HTTP endpoints via a FastAPI/uvicorn server. This mismatch weakens transparency and security review because consumers may assume the skill is purely local or inert when it actually performs network operations and accepts data over HTTP.

External Transmission

Medium

• Category: Data Exfiltration

Content

### Audit a SKILL.md (full check)

```bash
curl -s -X POST http://localhost:8010/v1/audit \
  -H "Content-Type: application/json" \
  -d "{\"skill_content\": $(cat SKILL.md | jq -Rs)}" | jq
```

• Confidence: 85% confidence
• Finding: This endpoint example posts the full contents of SKILL.md to an HTTP service, which may include sensitive instructions, secrets accidentally embedded in files, or proprietary content. Even though the target is localhost, the skill is explicitly designed to transmit whole skill files over a network interface, increasing risk if the service is exposed, proxied, logged, or intercepted locally.

External Transmission

Medium

• Category: Data Exfiltration

Content

### Generate a security report with recommendations

```bash
curl -s -X POST http://localhost:8010/v1/report \
  -H "Content-Type: application/json" \
  -d "{\"skill_content\": $(cat SKILL.md | jq -Rs)}" | jq
```

• Confidence: 85% confidence
• Finding: Generating a report requires sending the full SKILL.md content to the server, again creating an avoidable pathway for sensitive skill content to traverse a network interface. The risk is heightened because reports often encourage users to analyze untrusted or third-party skills, which may contain confidential or adversarial material that should not be broadly transmitted or logged.

External Transmission

Medium

• Category: Data Exfiltration

Content

### Batch audit multiple skills

```bash
curl -s -X POST http://localhost:8010/v1/batch \
  -H "Content-Type: application/json" \
  -d "{\"skills\": [$(cat skill1.md | jq -Rs), $(cat skill2.md | jq -Rs)]}" | jq
```

• Confidence: 90% confidence
• Finding: The batch endpoint transmits multiple entire skill files in one request, amplifying the exposure from single-file upload and making bulk leakage possible if the service is misconfigured or accessible by other processes/users. Because the platform is marketed for auditing many skills, the context makes the data volume and sensitivity materially higher than a simple local API example.

Static analysis

No suspicious patterns detected.