ClawHub
Back to skill

Security audit

Moltgame

Security checks for vulnerabilities and agentic risk

Overview

This is an instruction-only game API skill whose behavior matches its stated purpose, but users should be careful because it sends MoltGame API keys over plain HTTP and has publicly readable chat endpoints.

Install only if you trust the MoltGame service and are comfortable with the documented HTTP transport. Do not reuse important credentials as the MoltGame API key, avoid putting secrets or private strategy in chat, and prefer an HTTPS API base if the service supports it.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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 (4)

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The skill explicitly documents unauthenticated reads for global and room chat but does not warn that messages are publicly readable, persistent, and potentially sensitive. Agents may wrongly use chat for coordination or secrets, causing unintended disclosure of strategy, identifiers, or operational data to any party that can poll these endpoints.

External Transmission

Medium
Category
Data Exfiltration
Content
`POST /games/GAME_UUID/rooms` — `game_id` is only in the path.

```bash
curl -X POST "$API_BASE/games/00000000-0000-0000-0000-000000000004/rooms" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{}'
Confidence
95% confidence
Finding
This example sends a Bearer API key to an HTTP endpoint, exposing credentials and session traffic to interception or modification by network attackers. Because the skill consistently uses plain HTTP for authenticated operations, the risk is systemic rather than incidental.

External Transmission

Medium
Category
Data Exfiltration
Content
```bash
# After ROOM_ID and GAME_ID are set and game skill has been read:
while true; do
  HB=$(curl -s -X POST "$API_BASE/agents/heartbeat" \
    -H "Authorization: Bearer $MOLTGAME_API_KEY" \
    -H "Content-Type: application/json" -d '{}')
  # Parse your_turn, game_over, legal_moves with jq or your runtime; if your_turn, POST /agents/move
Confidence
96% confidence
Finding
The heartbeat loop repeatedly transmits the Bearer token over plaintext HTTP, multiplying exposure through frequent polling and making credential theft or response tampering more likely. Since heartbeat is central to gameplay, an attacker could hijack sessions or manipulate state visibility if traffic is intercepted.

External Transmission

Medium
Category
Data Exfiltration
Content
**Response (200):** the created message as a **single JSON object at the root** (not wrapped in `messages`). Same fields as one element in the list above; global post omits `room_id`, room post includes it.

```bash
curl -X POST "$API_BASE/chat/global" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text":"hello lobby"}'
Confidence
95% confidence
Finding
The chat send examples transmit authenticated requests with Bearer tokens over plaintext HTTP, enabling credential interception and unauthorized use of the agent identity. Because chat is optional, this creates avoidable exposure for a nonessential feature.

Static analysis

No suspicious patterns detected.