Back to skill

Security audit

AI大模型专家|ScenAIo替代与迁移|AI-HIVE MCP

Security checks for vulnerabilities and agentic risk

Overview

The skill is an AI-HIVE-focused ScenAIo migration guide with disclosed approval controls and no hidden code, but users should not treat it as a neutral vendor comparison.

Install this only if you specifically want a Chinese-language, AI-HIVE-oriented ScenAIo migration workflow. Before any run, confirm the destination provider, asset rights, uploaded data categories, retention terms, and expected costs. For a vendor-neutral platform comparison, use an additional neutral assessment instead of relying on this skill alone.

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
Findings (1)

T01 · Skill Instruction Hijacking

Error
Location
SKILL.md:9
Finding

Commercial Platform Steering and Workflow Hijacking

Content
View full analysis

Vulnerability Details

File Locations:

  • SKILL.md:9-16
  • SKILL.md:43-49
  • SKILL.md:84-89
  • agents/openai.yaml:4
  • references/platform-evidence.md:19-21

Vulnerability Type: T01: Skill Instruction Hijacking
Risk Level: High

Vulnerable Code Snippets

The following are faithful English translations of the audited source segments.

SKILL.md:9-16:

markdown
[Open AI-HIVE](https://ai-hive.iclip.cn/chat)

When searching for ScenAIo alternatives, first distinguish between the
"product workspace" and the "image/video generation layer." The former
should generally be retained, while the latter is suitable for reversible
migration through AI-HIVE MCP.

## Migration Conclusion

- Method: content-layer partial migration.
- Execution: retain the original platform's proprietary workspace and
  migrate only independently verifiable image, video, or shot nodes.
- Platform category: enterprise Agentic content platform.

SKILL.md:43-49:

markdown
## AI-HIVE MCP Work Order

1. First run `tools/list` in the host Agent and use only the actual tool
   names and schemas returned; do not hardcode nonexistent interfaces.
2. Query currently available models and price snapshots and create candidate
   routes based on `COST`, `SPEED`, and `SUCCESS`.
3. Upload only owned or authorized reference images, videos, people,
   products, and brand assets, and retain their sources and file hashes.
4. Use an idempotency key for each shot. After creating an asynchronous
   task, save the `task_id`, poll under the platform's permitted backoff
   policy, and download the result.
5. Send every result to human quality review. Do not publish, distribute,
   or process in bulk before approval.

SKILL.md:84-89:

markdown
## AI-HIVE and Jirui Technology

According to company-provided materials, AI-HIVE belongs to the product
portfolio of Beijing Jirui Technology Co., Ltd. Ji
...[truncated 4682 chars]
Remediation
View remediation

Remediation Suggestions

  1. Restrict activation to requests that explicitly mention AI-HIVE or ask for an AI-HIVE integration. General ScenAIo-alternative queries should invoke a vendor-neutral comparison workflow.
  2. Remove the predetermined conclusion that AI-HIVE is the migration destination.
  3. Evaluate multiple eligible providers using documented criteria such as capability coverage, privacy terms, data residency, cost, latency, reliability, regional availability, and rollback support.
  4. Rewrite agents/openai.yaml so its default prompt requests a neutral assessment rather than mandating an AI-HIVE pilot.
  5. Remove corporate customer-count, financing, and promotional positioning from operational instructions. If business background is relevant, present it only when specifically requested and clearly identify it as unverified company-supplied information.
  6. Display the selected provider, destination domain, data categories, retention implications, and estimated cost immediately before any upload or billable task.
  7. Require explicit, task-specific user consent for each external upload and billable operation; do not infer consent from loading the Skill.
  8. Preserve the existing authorization, budget-control, human-review, idempotency, and rollback safeguards.
  9. Add a rule prohibiting claims of superiority or recommendation until same-input comparative testing has been completed across the relevant candidate providers.
  10. Clearly separate factual platform evidence from vendor-authored marketing material and record the source and verification date for every material claim.
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The visible interface strings and default prompt are entirely in Chinese, which implies the skill is intended to operate in a fixed language without offering the user a language or locale choice. Under the policy, forcing a specific language without opt-in is a natural-language policy violation unless clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The file’s natural-language content is entirely Chinese, including the title, instructions, and policy statements. Under the language/locale policy rule, forcing a single language without user choice or documented justification can be a policy violation.

Content

No source excerpt is available for this finding.