Looking for Someone

Security checks across malware telemetry and agentic risk

Overview

This appears to be a purpose-aligned local missing-person case organizer, but it stores highly sensitive case details in unencrypted local JSON and has minor packaging/documentation gaps.

This skill looks coherent and local-only, but it may contain very private information about missing people and families. Use it on a trusted device, avoid storing unnecessary ID numbers or financial details, carefully review generated notices before sharing them publicly, and confirm how to delete local case data when it is no longer needed.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

#
ASI06: Memory and Context Poisoning
Medium
What this means

Anyone with access to the user’s local files or backups could potentially read sensitive case details if the device or account is not protected.

Why it was flagged

The skill intentionally stores sensitive missing-person case data locally, including possible ID numbers and family contact information, and discloses that it is not encrypted.

Skill content
Optional Fields: `phone`, `birthDate`, `idNumber`, ... `familyContacts`; Data Storage: Cases are saved locally in `~/.openclaw/skills-data/looking-for-someone/`; Limitations: No field-level encryption
Recommendation

Only enter sensitive identifiers when truly necessary, avoid putting ID numbers or private addresses into public notices, protect the local device account, and delete or archive case files when no longer needed.

#
ASI04: Agentic Supply Chain Vulnerabilities
Info
What this means

Users may not be able to run the documented tests as written, making it harder to verify the package locally.

Why it was flagged

The documented test command references test.js, but the provided file manifest does not include that file. This is a completeness/documentation gap, not evidence of hidden behavior.

Skill content
Run the basic test suite:
```bash
node test.js
```
Recommendation

Before relying on the skill, verify the available files and treat the missing test reference as a packaging issue to resolve with the publisher.

#
ASI09: Human-Agent Trust Exploitation
Low
What this means

Users could assume there is an obvious built-in deletion workflow even though the primary instructions do not show one.

Why it was flagged

The privacy document describes deletion rights, but the main SKILL.md command list only documents create, list, progress, clue, notice, guide, and reminders commands, so deletion mechanics are not clearly documented in the primary workflow.

Skill content
Deletion: User-initiated deletion available at any time
Recommendation

Confirm how to delete stored case data before entering highly sensitive information; if no command is available, delete the local data files from the documented storage directory when appropriate.