ClawHub
Back to skill

Security audit

RxnIM

Security checks for vulnerabilities and agentic risk

Overview

The skill does what it says, but users should review it because it sends reaction images to a third-party service and gives unpinned local install commands that execute external code.

Install or use this only if you are comfortable sending reaction images to the named HuggingFace Space. Do not use the hosted path for unpublished, proprietary, regulated, or confidential chemistry data unless you have approval. If using local deployment, pin and review the repository, dependencies, model archive, and eval script, and run them in an isolated environment without unnecessary credentials or sensitive mounts.

A.I.G

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:74
Finding

Unpinned Remote Dependencies and Script Execution

Content
View full analysis
Remediation
View remediation

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill recommends sending base64-encoded reaction images to a third-party HuggingFace Space but does not clearly warn users that uploaded content leaves the local environment. Reaction schemes may contain unpublished research, proprietary synthesis routes, or other confidential data, so silent transmission creates a real privacy and compliance risk.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
91% confidence
Finding

The sample code performs an outbound POST containing the full image payload to an external endpoint. This is expected for a hosted inference service, but it is still a genuine external transmission path that can expose sensitive scientific or business information if used on confidential images without safeguards.

Content

Scanner excerpt · SKILL.md (reported line 57)May include surrounding context.

md
def parse_reaction_image(image_path: str, hf_space: str = "https://CYF200127-RxnIM.hf.space") -> dict:
    with open(image_path, "rb") as f:
        b64 = base64.b64encode(f.read()).decode()
    resp = requests.post(
        f"{hf_space}/run/predict",
        json={"data": [f"data:image/png;base64,{b64}"]},
        timeout=120

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The workflow explicitly says images may be obtained from a user upload, URL, or file path, which expands the skill from image parsing into network retrieval of arbitrary resources. If an implementation follows this guidance without strict allowlists and validation, it can enable server-side request forgery, access to internal-only endpoints, or retrieval of sensitive content under the guise of image processing.

Content

No source excerpt is available for this finding.