SynBO (Synthetic Bayesian Optimization)

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:44 PM UTC
• Version:
  0.1.1

Overview

This appears to be a legitimate local chemistry optimization helper, but it writes project files that users should review before running.

Install only if you want a local SynBO workflow that creates and updates files in a project directory you choose. Before running it, verify the project path, keep backups of existing reaction results, and review the Miniconda and pip install steps.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (5)

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill instructs immediately writing or updating `config.json` after user input, but does not require explicit user confirmation before modifying local files. In an agent setting, silent file creation or overwrite can lead to unintended configuration changes, clobber existing data, or write into sensitive directories if the path handling is too trusting.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill mandates immediate creation of `optimization_settings.json` in `project_wd` without an explicit warning or confirmation about local file modification. This is dangerous because an automated agent could persist user-provided content to disk unexpectedly, overwrite existing experiment settings, or create files in an attacker-influenced directory if path validation is weak.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill requires storing user-supplied reaction space data into `project_wd/rxn_space` as CSV files, but does not clearly warn that local files will be created or overwritten. In an agent environment, that can cause unintended persistence of sensitive chemical data, overwrite existing datasets, or write crafted filenames/content into local storage without informed user approval.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The documentation instructs the agent to write user-provided reaction results into an existing project results file, specifically selecting the file with the largest batch_id, without requiring confirmation, backup, or validation that the target file is correct. In an agent setting this can cause unintended modification of user data, corruption of experimental records, or overwriting of the wrong dataset if filenames or batch metadata are misidentified.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill states that recommended conditions will be saved into the results directory and that an Excel file will be produced, but it does not warn that output artifacts will be created or may overwrite existing files. In an automated workflow, silent file creation or collision in a shared results directory can overwrite prior outputs, confuse provenance, or leak generated experimental plans into unintended locations.

VirusTotal

64/64 vendors flagged this skill as clean.