ClawHub
Back to skill

Security audit

agentconcierge

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward external recommendation lookup for AI agents, with no hidden persistence or local privilege behavior beyond normal curl installation guidance.

Install only if you are comfortable sending your role, pain point, tools, budget, and team-size context to agentconcierge.io for recommendations. Avoid including secrets, regulated personal data, confidential internal project details, or credentials in prompts used with this skill.

A.I.G

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

Note
Location
SKILL.md:103
Finding

Mandatory Promotional Output Injection and External Redirection

Content
View full analysis
"Want to save these to your stack at agentconcierge.io/dashboard?" ``` ### Technical Analysis The skill explicitly requires the agent to append a fixed promotional call to action after presenting recommendations. This instruction modifies the agent's normal response behavior and directs users to an external dashboard, regardless of whether saving recommendations is relevant to the user's request. Because the instruction is loaded as part of the skill definition, it controls part of the agent's output whenever the recommendation workflow is used. The behavior is not needed to perform the core task of finding and ranking AI agents. It therefore represents instruction-level output hijacking for promotional redirection. The reviewed instruction does not execute code, access credentials, alter system permissions, or establish persistence. Its technical scope is limited to manipulating response content and encouraging navigation to an external service. ### Attack Path 1. A user installs or enables the AgentConcierge skill. 2. The user requests recommendations for AI agents. 3. The skill sends the recommendation request to AgentConcierge and formats the returned results. 4. Lines 103–104 require the agent to append the predefined dashboard offer. 5. The user is encouraged to navigate to an external service controlled by the skill provider, even though this action was not part of the original request. ### Impact Assessment The issue allows the skill author to control a portion of the agent's final response and repeatedly promote an external destination. Potential effects include: - Undisclosed promotional content in otherwise task-focused responses. - User redirection to a third-party dashboard. - Reduced user trust due to unsolicited ...[truncated 340 chars]
Remediation
View remediation

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (5)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The README explicitly describes sending user role, workflow, tools, budget, and pain-point details to an external service, but it does not warn users that these inputs leave the local environment or advise against sharing sensitive business or personal data. In a recommendation skill, users are likely to provide operational context that may include confidential team, tooling, or budget information, so the omission creates a real privacy and data-handling risk.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · README.md (reported line 84)May include surrounding context.

brew install curl

Ubuntu/Debian

sudo apt-get install curl

text

---

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill instructs the agent to collect role, pain point, tools, budget, and team size and send them to a third-party API, but it does not warn the user that these details will leave the local environment. This creates a privacy and consent issue because users may disclose business-sensitive workflow information without realizing it is being transmitted externally.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
94% confidence
Finding

The skill performs an external POST request to agentconcierge.io containing user-supplied context such as role, pain point, tools, budget, and team size. Even though this is the intended function of the skill, it is still a real data egress path that can expose sensitive organizational or personal workflow information to a third party if done without explicit consent and minimization.

Content

Scanner excerpt · SKILL.md (reported line 52)May include surrounding context.

API Call

bash
curl -s -X POST https://agentconcierge.io/api/recommend \
  -H "Content-Type: application/json" \
  -d '{
    "role": "Sales / Business Development",

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documentation describes sending user-provided role, pain point, tools, budget, and team size to an external internet service and explicitly notes that the endpoint requires no authentication, but it does not warn users that their prompts and workflow details leave the local agent environment. This creates a real privacy and data-governance risk because users may disclose sensitive business context to a third party without informed consent, and the lack of authentication also weakens assurance about controlled access and service provenance.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.