Back to skill

Security audit

AI Adoption Audit

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent company AI research/reporting skill, with the main caution that it saves and reuses local report archives by default.

Install only if you are comfortable with the agent saving company AI-audit reports locally and reusing recent reports for follow-up requests. Avoid including confidential customer notes or sensitive internal company context unless you are comfortable with that content being written into the report archive.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The manifest description includes many broad trigger terms such as '企业AI', 'ai adoption', and '公司AI调研' without clear boundaries. Overbroad activation can cause the skill to run on loosely related prompts, leading to unintended web research, file access, or storage behaviors when the user did not specifically request this skill.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill instructs the agent to persist company reports locally, reuse prior outputs, and perform incremental updates based on stored history. This expands behavior from a one-shot research task into local stateful storage, which can retain user-provided company context and create unannounced data persistence, stale-data reuse, and cross-session leakage risks.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill describes writing reports to local paths, replacing prior files, and maintaining history, but does not warn the user or request consent for local persistence. Silent local writes are dangerous because they can store sensitive user-supplied notes, create unexpected artifacts on disk, and enable later disclosure or tampering through reused cached reports.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
79% confidence
Finding

The example trigger phrase '帮我调查一下 XX 公司的 AI 落地情况' is underspecified and may match common information-seeking requests that do not imply consent for the skill's full workflow. In this skill, that matters more because the workflow includes multi-round external research and possible report reuse/storage.

Content

No source excerpt is available for this finding.