Security audit

doubao-opencli

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed Doubao automation skill, but it controls a logged-in browser session and has an unsafe PPT input path plus sensitive local logging/backups that users should review carefully.

Install only if you are comfortable letting scripts automate your logged-in Doubao account through Edge. Use a separate browser profile or account if possible, do not feed untrusted text or files into the PPT topic/outline/draft flow until the eval handling is fixed, and regularly review or delete local output, logs, and backup files that may contain private prompts, documents, or conversations.

SkillSpector

By NVIDIA

Vulnerability Patterns

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

Findings (7)

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The script persistently logs the user's prompt and generated image URLs to a local markdown file without any consent, warning, or opt-in. Prompts may contain sensitive business data, personal information, or confidential creative content, and local log files can later be accessed by other users, backup systems, or malware on the host.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The script writes a preview of user-supplied input to a plaintext log file, including topic, outline, or draft content. Because this skill is designed to process arbitrary user documents and prompts, the logged preview can expose sensitive business, personal, or proprietary data to anyone with access to the filesystem or log collection pipeline.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: When a draft file is provided, the script reads its full contents and injects them into a browser session that submits the text to doubao.com for PPT generation. In context this data transfer is core functionality rather than covert exfiltration, but it is still a real privacy/security issue if users are not clearly warned that local document contents will be sent to an external AI service.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The backup command exports the full conversation history and detailed conversation contents to a local JSON file without an explicit warning, consent prompt, or access-control guidance. In this skill's context, conversations may contain sensitive prompts, generated content, personal information, or business data, so writing everything to disk increases the risk of accidental disclosure through shared machines, backups, or improper file permissions.

Missing User Warnings

Medium

Confidence: 84% confidence
Finding: The meeting summary function automatically saves summary data derived from conversations to a local file after fetching it, without clearly informing the user that content will be persisted. Meeting summaries often contain sensitive business discussions, action items, or personal data, so silent local persistence can create confidentiality and compliance issues.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: Batch mode reads prompts from a file, sends them to the external Doubao service, and then writes responses to a local markdown file without an explicit privacy warning. This creates two exposure paths—outbound transmission of potentially sensitive file contents and local persistence of model outputs—making accidental leakage more likely in automation or shared environments.

Ssd 3

Medium

Confidence: 97% confidence
Finding: The script logs user-provided outline or draft-derived prompt text in plaintext to a timestamped file under the output directory. Since this automation is intended for batch processing and backups, logs may accumulate sensitive internal content over time and be copied, synced, or exposed more broadly than the original input.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.