Trusted — Risk Score 0/100
Last scan:2 days ago Rescan
0 /100
clawmate
AI boyfriend/girlfriend companion with 8 personas, relationship stages, proactive messaging, and shared memory
ClawMate is a benign AI companion skill with no executable code, network calls, or sensitive operations. It consists solely of markdown persona definitions and JSON memory files for relationship simulation.
Skill Nameclawmate
Duration39.5s
Enginepi
Safe to install
This skill is safe to use. No security concerns identified.
ResourceDeclaredInferredStatusEvidence
Filesystem READ READ ✓ Aligned SKILL.md declares reading persona and memory files
Filesystem WRITE WRITE ✓ Aligned SKILL.md section 9 Memory Protocol declares writing to memory JSON files
Network NONE NONE SKILL.md Privacy section explicitly states 'No data is sent to external services…
Shell NONE NONE No shell scripts or subprocess calls in codebase
Environment NONE NONE SKILL.md does not access os.environ
Skill Invoke NONE NONE No skill_invoke usage found
Clipboard NONE NONE No clipboard access in codebase
Browser NONE NONE No browser automation in codebase
Database NONE NONE No database access in codebase
1 findings
📧
Info Email 邮箱地址
[email protected]
SKILL.md:456

File Tree

13 files · 124.1 KB · 2394 lines
Markdown 10f · 2328L JSON 3f · 66L
├─ 📁 memory
│ ├─ 📋 message_pool.json JSON 15L · 210 B
│ ├─ 📋 shared_memories.json JSON 18L · 478 B
│ └─ 📋 user_profile.json JSON 33L · 984 B
├─ 📁 personas
│ ├─ 📝 cheerful.md Markdown 181L · 9.7 KB
│ ├─ 📝 chill.md Markdown 192L · 10.7 KB
│ ├─ 📝 cool.md Markdown 200L · 8.2 KB
│ ├─ 📝 dominant.md Markdown 191L · 9.8 KB
│ ├─ 📝 gentle.md Markdown 160L · 7.0 KB
│ ├─ 📝 intellectual.md Markdown 183L · 11.6 KB
│ ├─ 📝 playful-dark.md Markdown 189L · 11.1 KB
│ └─ 📝 tsundere.md Markdown 177L · 9.4 KB
├─ 📝 relationship.md Markdown 159L · 6.2 KB
└─ 📝 SKILL.md Markdown 696L · 38.7 KB

Security Positives

✓ No executable code - skill consists entirely of Markdown and JSON files
✓ Explicit privacy statement: 'No data is sent to external services'
✓ All file operations are declared and scoped to skill's memory directory
✓ No credential harvesting or environment variable access
✓ No shell execution, subprocess, or system command calls
✓ No obfuscated code, base64 payloads, or hidden functionality
✓ WeChat OpenID ([email protected]) is a delivery target parameter, not data exfiltration
✓ User consent flow required for proactive messaging setup
✓ Data deletion and export commands give users full control
✓ No suspicious network patterns (direct IPs, curl, wget)