deso-research

• Outcome: Pass
• Latest audit: May 28, 2026 at 11:35 AM UTC
• Version:
  1.0.0

Overview

This is a coherent read-only social-research skill, with ordinary cautions around installing a third-party CLI and using optional social-network credentials.

Install only if you trust the deso-ag npm package, ideally in an isolated environment if you are cautious. Provide only the API keys or Bluesky app password needed for the networks you want to query, and avoid putting secrets or sensitive personal details into searches because queries may be sent to external social-network services.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• 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 (3)

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The trigger guidance is unusually broad and repeatedly instructs the agent to use this skill for essentially any decentralized-social query, including vague prompts like 'check Farcaster'. Over-broad routing can cause unintended external data access, unnecessary tool execution, and mismatched handling of user intent, especially when a simpler answer or clarification would be safer.

Missing User Warnings

Medium

• Confidence: 84% confidence
• Finding: The skill directs the agent to query external social platforms and aggregate results, but it does not warn that user queries may be sent to third-party services or that returned content may contain sensitive, personal, or misleading information. This omission increases privacy and transparency risk because users may not realize their prompt is being used to access outside services and process public social data across multiple networks.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The documentation explicitly encourages use of a compact format containing full untruncated post content and describes Bluesky search using app-password-based authentication, but provides no guidance on safe handling, minimization, retention, or redaction of potentially sensitive user-generated content and credentials. In a research skill intended for broad social-content analysis, this increases the chance that downstream agents ingest, store, summarize, or expose personal data and account secrets without adequate safeguards.

VirusTotal

65/65 vendors flagged this skill as clean.