RenderMark

• Outcome: Pass
• Latest audit: Jul 31, 2026 · 4:24 PM UTC
• Version: 0.1.10

Overview

This is a coherent document rendering and publishing skill, but users should treat publish/share actions and stored credentials as sensitive.

Install only if you want an agent to send markdown to RenderMark, Google Docs, or GitHub-backed workflows and to manage documents in that account. Review content before publishing or sharing, avoid syncing private repository material with publish enabled unless intended, and protect ~/.rendermark/config.json because it may contain API keys or OAuth tokens.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Rogue Agent: Self-Modification, Session Persistence
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (7)

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill instructs the agent to publish documents to rendermark.app and share them with other people, but it does not require an explicit user-facing warning or confirmation that the document contents will be transmitted to a third-party service. This creates a real risk of unintended exfiltration of sensitive notes, internal documents, or proprietary content when a user asks to 'share' or 'publish' without appreciating the external transfer.

Missing User Warnings

Low

• Confidence: 89% confidence
• Finding: The GitHub sync flow tells the agent to import content from an external repository and, in examples, potentially publish it, but it does not warn the user that external content is being fetched and may be republished. This can lead to accidental ingestion of untrusted content or unintended publication of repository material, especially when users assume a local-only transformation workflow.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The setup guide explicitly instructs users to place a live API key in a plaintext file under their home directory, but provides no guidance on restricting file permissions, avoiding accidental inclusion in backups/repos, or preferring safer secret storage. If that file is exposed through local compromise, misconfiguration, syncing, or logging, the credential could be used to access the RenderMark account and associated document operations.

Missing User Warnings

Medium

• Confidence: 87% confidence
• Finding: The guide tells users to place Google OAuth client credentials in the same local config file without any warning about protecting those secrets. While OAuth client secrets are sometimes treated as lower sensitivity than user tokens, exposing them can still enable abuse of the app configuration, phishing-like impersonation in some contexts, or unauthorized use of the integration setup.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The publishing and sharing tools send user-supplied markdown to external services such as rendermark.app and Google Docs, but the reference does not warn that document contents will leave the local environment. In an agent setting, this can cause unintended disclosure of sensitive data because a user may interpret 'publish' or 'share preview' as a formatting action rather than an outbound data transfer.

Missing User Warnings

Low

• Confidence: 82% confidence
• Finding: The export tools write rendered output to disk, including to a caller-specified outputPath, but the reference does not call out that files will be created locally. In an agent workflow this can lead to unexpected persistence of sensitive content, accidental overwrites, or writing into shared/synced directories without the user realizing it.

Session Persistence

Medium

• Category: Rogue Agent

Content

- `title` (string, required): Document title

### share_live_preview
Create a temporary shareable preview URL.

**Parameters:**
- `markdown` (string, required): Markdown content

• Confidence: 73% confidence
• Finding: The live preview feature creates a shareable URL that persists for a selectable lifetime up to 7 days, which introduces temporary but real access persistence outside the original session. If used with sensitive markdown, anyone possessing the link may access the content during that window, and the reference does not emphasize this exposure.

VirusTotal

63/63 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.