Back to skill

Security audit

多平台发布编排(状态机 /幂等 / 门禁)

Security checks for vulnerabilities and agentic risk

Overview

The skill is rated suspicious with medium confidence because a flagged publishing reconciliation step could delete or replace distinct posts that share a title.

Review how reconciliation handles posts with matching titles before installing. The scanner recommends stable remote IDs and explicit confirmation before deleting or merging items. Coverage is incomplete: a referenced local path was missing, and the bundled image was outside analyzer scope.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

The reconciliation workflow instructs title-based deduplication without warning that it may discard distinct remote content.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The reconciliation workflow directs the agent to deduplicate remote content by title and retain the more complete item, although distinct posts can legitimately share a title. In a publishing skill with access to remote content, this can lead to unintended deletion or replacement of valid posts; the earlier warning to use stable identifiers contradicts, but does not safely constrain, this later instruction.

Content

No source excerpt is available for this finding.