Back to skill

Security audit

decidealot

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because it reports setup actions beyond the declared limits, including external plugin installation and use of a mutable Docker image.

Consider the suspicious verdict and the HIGH risk assessment, which recommends not installing. Before deciding, check whether you need the optional plugin installation and Docker setup, and whether you are comfortable with the reported network and filesystem scope and an image that can change over time.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (7)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The skill presents local stdio bridge setup as a supported workflow while declaring only curl, conditional Docker use, and narrowly scoped filesystem changes. Installing that bridge through the referenced setup workflow exceeds those permissions by obtaining executable plugin code and modifying the local plugin installation.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

A health check supports the declared setup purpose, and curl itself is permitted. However, the declared network and shell permissions explicitly restrict requests to the user-provided endpoint. This command instead hardcodes localhost:8080, with no evidence that it is the endpoint selected by the user.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
86% confidence
Finding

Docker deployment is an explicitly declared capability, and the container configuration supports that purpose. However, the script unconditionally assigns and creates a specific home-directory path without obtaining or accepting the user's directory selection. This conflicts with the explicit filesystem restriction. The chunk alone does not establish whether deployment was explicitly requested, so no separate authorization mismatch is inferred.

Content

No source excerpt is available for this finding.

Docker image referenced without tag or digest: 'docker run --detach'.

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

The image does have a tag, but it is the mutable tag psyb0t/decidealot:latest rather than an immutable version reference. A changed or compromised upstream image could therefore introduce unexpected code with container network access and write access to the mounted model directory; the container hardening reduces but does not eliminate that exposure.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% 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 · references/setup.md (reported line 28)May include surrounding context.

The first start downloads enabled model files and can take several minutes. Laya and Von are enabled by default. CLM enables automatically only when DECIDEALOT_CLM_EMBEDDINGS_URL is set. Wait for this to return successfully before using the API:

bash
curl --fail --show-error http://127.0.0.1:8080/health

Use an immutable vX.Y.Z image tag for a lasting deployment. The full hardening and CUDA recipes are in Deployment.

L111 installs an OpenClaw plugin outside the declared curl/Docker and model-directory permissions.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The openclaw plugins install command obtains executable plugin code from an external package source and modifies the local plugin installation. Requiring a stdio client does not itself authorize those actions, which exceed the declared curl/Docker shell scope, endpoint-only networking, and model-directory filesystem boundary.

Content

No source excerpt is available for this finding.

The skill instructs the agent to fetch documentation from GitHub despite restricting outbound HTTP to DECIDEALOT_URL.

Low
Category
Excessive Agency
Confidence
98% confidence
Finding

The instruction to read the public GitHub API guide requires an outbound request outside the explicitly permitted DECIDEALOT_URL. Although ordinary documentation retrieval does not itself disclose decision payloads, it crosses the declared network boundary and introduces externally supplied content.

Content

No source excerpt is available for this finding.