Back to skill

Security audit

tech-research

Security checks for vulnerabilities and agentic risk

Overview

Scanners returned a benign verdict with medium confidence: the findings mostly fit a technical research skill, though broad activation, external search privacy, and incomplete reference checks leave some uncertainty.

Before installing, consider whether the broad research trigger fits your needs and whether you may use the skill with confidential research context. The scan was partial: two referenced memory paths were missing, so their role could not be fully assessed.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

The standalone keyword “调研” activates the skill without requiring a technical or cloud-phone research context.

Medium
Category
Trigger Abuse
Confidence
96% confidence
Finding

The trigger activates whenever the user mentions listed keywords, including the generic term “调研” (research). Although the introduction describes cloud-phone technical research, the activation rule does not enforce that scope or exclude unrelated research requests and incidental mentions.

Content

No source excerpt is available for this finding.

The external-search workflow lacks a privacy warning about transmitting research queries to third-party services.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The workflow retrieves internal knowledge-base material and then conducts external searches without a rule separating confidential context from public search terms. Research questions or query terms could therefore disclose internal project names, architectural details, or proprietary plans to search providers; actual disclosure is not demonstrated by the supplied content.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · assets/radar_sources.json (reported line 26)May include surrounding context.

json
"keywords": ["WebRTC", "release", "new features"],
      "fallback": {
        "type": "github",
        "url": "https://api.github.com/repos/chromium/chromium/commits?path=third_party%2Fwebrtc&per_page=30",
        "note": "Chromium third_party/webrtc autoroll commits,间接跟踪 WebRTC 演进(2026-11-09 雷达已验证可用)"
      }
    },

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · assets/radar_sources.json (reported line 37)May include surrounding context.

json
"keywords": ["Chromium", "release", "WebRTC"],
      "fallback": {
        "type": "github",
        "url": "https://api.github.com/repos/chromium/chromium/tags?per_page=20",
        "note": "Chromium 版本标签,跟踪发布节奏"
      }
    },

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · assets/radar_sources.json (reported line 55)May include surrounding context.

json
{
      "name": "FFmpeg Release",
      "type": "github",
      "url": "https://api.github.com/repos/FFmpeg/FFmpeg/tags?per_page=10",
      "keywords": ["release", "codec"],
      "fallback": null
    },

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · assets/radar_sources.json (reported line 73)May include surrounding context.

json
{
      "name": "GitHub 云手机/云游戏新仓库",
      "type": "github_search",
      "url": "https://api.github.com/search/repositories?q=cloud+phone+OR+cloud+gaming+android+created%3A%3E%7Bsince%7D&sort=stars&per_page=10",
      "keywords": ["cloud phone", "cloud gaming", "waydroid", "moonlight", "sunshine"],
      "fallback": null
    }