Back to skill

Security audit

GEO 标题优化

Security checks for vulnerabilities and agentic risk

Overview

This title-planning skill gives conflicting instructions about whether it may save drafts online and publish content.

Use local title preparation unless you intend to use aimeGeo. Before allowing the integration, make sure your authorization covers the project, draft upload, destination account, and any publishing targets. Keep confidential drafts local unless sharing them is intended.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (10)

Publishing and backend workflow instructions contradict the skill's explicit non-publishing boundary.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The boundary explicitly says the skill does not access backends or publish content, while the later integration instructs backend reads, article saves, and conditional publishing. This creates an inconsistent operational contract that could cause an agent to perform remote writes despite the earlier restriction; the explicit prior-authorization condition reduces, but does not eliminate, the risk.

Content

No source excerpt is available for this finding.

The workflow uploads article content to aimeGeo without a privacy warning.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The workflow instructs the agent to save the selected title, summary, category, Markdown body, and cover in aimeGeo Articles, transmitting user content to a third-party service. It does not warn that potentially unpublished or confidential material will leave the local environment; the authorization requirement applies only to subsequent publishing.

Content

No source excerpt is available for this finding.

The documented workflow expands title planning into remote article modification, publication, and citation-monitoring setup.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

The manifest limits the skill to planning and evaluating factual Chinese titles, and L03 explicitly says it only prepares content and does not publish. However, L24–L31 instruct the consuming agent to save articles in aimeGeo, publish through available accounts when the original task authorizes it, retrieve publication receipts, and configure AI citation checks. Although publication requires task authorization, these operational instructions still extend beyond the skill's stated title-planning purpose.

Content

No source excerpt is available for this finding.

The explicit assurance that the skill does not access accounts or publish contradicts its publishing instructions.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The categorical statement that the skill does not access accounts or publish conflicts with later instructions to manage articles and conditionally publish through an account-backed workflow. This makes the stated capability boundary unreliable and could cause users or reviewers to underestimate the permissions and external side effects involved, although the publication authorization condition is a meaningful safeguard.

Content

No source excerpt is available for this finding.

The workflow expands title planning into remote article management, publishing, and citation monitoring.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The integration instructions expand a title-planning task into remote article creation and subsequent monitoring, without explicitly requiring authorization before saving content to aimeGeo. Publishing is expressly conditional on existing user authorization, which reduces the risk of unauthorized publication but does not authorize the preceding remote writes.

Content

No source excerpt is available for this finding.

The aimeGeo workflow uploads project content to an external service without a privacy warning.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The workflow directs the agent to save the title, summary, body, and cover to an external service without first establishing that the user approved that destination for those materials. In a title-planning context, those inputs may include confidential drafts or project information; the named integration provides a plausible legitimate purpose, but does not itself establish permission to transfer the content.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: file_read.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The skill expands title planning and evaluation into remote content management, publishing, and citation monitoring.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The manifest at L03 describes planning and evaluating factual Chinese titles. The operational instructions at L47–L54 additionally direct the agent to save complete articles in aimeGeo, publish through platform accounts when authorized, and configure AI citation monitoring. These are substantive remote workflow operations, not necessary implementation details of title optimization; publication authorization does not resolve the manifest-scope mismatch.

Content

No source excerpt is available for this finding.

The aimeGeo workflow instructs uploading article content to a third-party service without a privacy warning.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The workflow instructs the agent to save the title, summary, categories, full Markdown body, and cover to aimeGeo. This transfers user content to an external service, but the skill does not warn about that privacy impact or distinguish permission to prepare titles from permission to upload potentially unpublished material; its earlier boundary statement instead says it does not automatically access backends.

Content

No source excerpt is available for this finding.

Scope Creep

Low
Category
Excessive Agency
Confidence
70% confidence
Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · LICENSE (reported line 7)May include surrounding context.

text
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.