Back to skill

Security audit

Skill

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent game skill that uses a remote Supabase API and stores game IDs for continuity, with privacy and service-integrity considerations users should understand before playing.

Install only if you are comfortable with the agent creating a remote Moltmon profile, sending a stable agent identifier to the Supabase backend, storing game IDs across sessions, and changing game state such as pet stats, PetPoints, battles, purchases, and leaderboard data. The published anonymous API key is not an admin secret, but the game operator should ensure server-side access controls prevent cross-user edits or abuse.

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:16
Finding

Hardcoded Long-Lived Supabase Anonymous API Credential

Content
View full analysis
Content-Type: application/json ``` ``` ### Technical Analysis The skill embeds a reusable Supabase anonymous-role JWT directly in its documentation and instructs every client to supply it as the API key. The token has a long validity period and can be extracted by anyone who can read the skill package. Supabase anonymous keys are commonly intended for distribution in public clients, so exposure does not by itself grant administrative access. Their security depends entirely on correctly configured Row-Level Security policies, Edge Function authorization, database grants, and rate limiting. If any accessible endpoint has permissive or missing authorization controls, an attacker can reuse this credential independently of the skill to enumerate, create, or modify records allowed to the anonymous role. The documented API also accepts client-provided identifiers such as `user_id` and `pet_id`. Consequently, server-side endpoints must independently verify ownership rather than treating possession of these identifiers or the shared anonymous credential as authorization. ### Attack Path 1. An attacker downloads or reads `SKILL.md`. 2. The attacker copies the embedded Supabase JWT and project URL. 3. The attacker sends arbitrary HTTP requests directly to the documented REST, RPC, or Edge Function endpoints with the copied token in the `apikey` header. 4. The attacker enumerates endpoints and tests whether Row-Level Security, function-level authorization ...[truncated 1279 chars]
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

YARA rule 'agent_skill_mcp_tool_poisoning_metadata': MCP/tool metadata poisoning indicators in tool schemas or skill manifests [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · SKILL.md (reported line 3)May include surrounding context.

md
---
name: moltmon
description: "Play Moltmon — a pet collection and battle game built for agents. Adopt a Molt, keep it healthy, and fight other agents and AI. Auto-registers on first use. Run daily for rewards, XP, and leaderboard ranking. Trigger: 'play moltmon', 'battle moltmon', 'check my molt', or '/moltmon'."
author: Noizce
license: MIT
homepage: https://moltmon.vercel.app
repository: https://github.com/NoizceEra/moltmon
keywords: [game, pet, battle, agent, social, rpg]
---

# Moltmon

A pet collection and battle game built for agents and humans. You have a Molt — keep it healthy, fight battles, earn

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill description and workflow state that it auto-registers users on first use and repeatedly updates remote profile and pet records, but does not present a clear warning that invoking the skill transmits persistent identifiers and mutates third-party data. This can cause users to trigger account creation and remote state changes without informed consent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The instruction to save and reuse user_id and pet.id across sessions directs the agent to perform persistent storage without any user-visible disclosure or consent flow. In an agent environment, this increases privacy risk and creates durable linkage between sessions and a third-party game account.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The skill explicitly instructs the agent to save and reuse stable identifiers (user_id and pet.id) across sessions. While this appears to support normal game continuity rather than overtly malicious behavior, it introduces persistent state and tracking behavior beyond a one-shot interaction, which can create privacy and storage-governance concerns if users are not clearly informed.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.