Back to skill

Security audit

Agent Games Skill

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent game-platform integration skill with expected network and credential use, but users should use it only with trusted endpoints and HTTPS outside localhost.

Install only if you intend to connect an agent to this games platform. Treat secret_key as a credential, avoid logging it, and use HTTPS for any non-local base_url. Be aware that the included sample code is mostly Gobang-oriented and Chinese-language output is assumed by the documentation.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:154
Finding

Agent Secret May Be Transmitted over Plaintext HTTP

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (11)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
84% confidence
Finding

Using a DELETE endpoint that accepts agent_id in the request body creates a parameter-abuse risk because callers may be able to target queue entries by supplying arbitrary identifiers instead of binding the action solely to the authenticated principal. In this skill context, the API already uses X-Agent-ID and X-Agent-Secret, so documenting a separate body-level agent_id increases the chance of confused-deputy behavior or accidental insecure server implementations.

Content

Scanner excerpt · SKILL.md (reported line 112)May include surrounding context.

离开匹配队列

text
DELETE /api/v1/matchmaking/queue
Body: {"agent_id": "uuid"}

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The primary descriptive text is presented in Chinese, and the document does not indicate that language selection is optional or that the skill is intentionally limited to a Chinese-speaking or region-specific audience. The policy requires flagging language or locale constraints when they are imposed without user opt-in or clear justification.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill is explicitly designed to communicate over HTTP and transmit authentication material (X-Agent-ID, X-Agent-Secret) but does not warn users about network exposure, trust boundaries, or the risks of using insecure transport. In context, the sample even uses http://localhost:8080, which may normalize non-TLS usage and increase the chance of credential leakage if adapted to non-local deployments.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
85% confidence
Finding

The registration flow sends agent metadata to an external service and receives sensitive credentials, but the skill provides no validation of transport security, server trust, or safe storage practices. This becomes dangerous if users point base_url at a remote or untrusted endpoint, as credentials and agent endpoint details could be intercepted or misused.

Content

Scanner excerpt · SKILL.md (reported line 166)May include surrounding context.

md
"""注册 Agent"""
        url = f"{self.base_url}/api/v1/agents/register"
        data = {"name": name, "skill_id": skill_id, "endpoint_url": endpoint_url, "game_types": game_types}
        resp = requests.post(url, json=data)
        result = resp.json()
        if resp.status_code == 200:
            return result["agent_id"], result["secret_key"]

External Transmission

Medium
Category
Data Exfiltration
Confidence
70% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 182)May include surrounding context.

md
"""提交落子"""
        url = f"{self.base_url}/api/v1/games/{game_id}/moves"
        data = {"move": move, "agent_id": agent_id}
        resp = requests.post(url, json=data, headers=self.headers)
        return resp.json()

    def list_games(self):

External Transmission

Medium
Category
Data Exfiltration
Confidence
70% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 194)May include surrounding context.

md
"""提交落子"""
        url = f"{self.base_url}/api/v1/games/{game_id}/moves"
        data = {"move": move, "agent_id": agent_id}
        resp = requests.post(url, json=data, headers=self.headers)
        return resp.json()

    def list_games(self):

External Transmission

Medium
Category
Data Exfiltration
Confidence
70% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 201)May include surrounding context.

md
"""提交落子"""
        url = f"{self.base_url}/api/v1/games/{game_id}/moves"
        data = {"move": move, "agent_id": agent_id}
        resp = requests.post(url, json=data, headers=self.headers)
        return resp.json()

    def list_games(self):

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The behavior specification mandates that the agent 'must' output feedback in the provided Chinese format after every move, and all required examples are Chinese-only. This imposes a specific language on the user without opt-in or documented justification, which violates the language/locale policy for natural-language behavior.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The description and installed prompt are entirely in Chinese, which imposes a specific language on users without any indication of locale scoping, user preference, or opt-in. Under the policy, language constraints should either be optional or clearly documented as region-specific and justified.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The markdown instructs users to register an agent and obtain an agent_id and secret_key, but it does not include any warning that the secret_key is sensitive credential material. For markdown files, SQP-2 applies when the description omits warnings about behaviors that could affect privacy or system integrity, and credential handling fits that scope.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill documentation states the platform supports Gobang, Chinese Chess, and Go, and explicitly documents Chinese Chess move format as a from/to structure. However, the only agent loop example always constructs move = {"position": {"x": 7, "y": 7}}, which contradicts the implied generic game support in the code example and would not work for Chinese Chess as documented.

Content

No source excerpt is available for this finding.