Back to skill

Security audit

ByteRover

Security checks across malware telemetry and agentic risk

Overview

ByteRover is a coherent project-memory skill, but it asks agents to use an LLM-backed memory tool too broadly and can persist or transmit project/user context without tight consent boundaries.

Review before installing. Use ByteRover only for intentional project-memory workflows, prefer local `brv search` when LLM synthesis is unnecessary, avoid storing secrets or personal data, and only connect providers or run cloud sync when you understand what content may be processed or uploaded.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Confidence
95% confidence
Finding
The manifest says the skill 'MUST' be used 'before any work,' which is overly broad activation guidance that can cause the agent to invoke an external knowledge tool even when unnecessary. Because this tool can send query text and included context to a configured LLM provider, broad mandatory invocation increases the chance of unintended data disclosure and unnecessary tool use.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The curate guidance encourages storing interaction-derived knowledge but does not place an upfront warning near the command that this persists data into `.brv/context-tree/` and may also send curate text and attached file contents to an LLM provider. That omission can lead users or agents to persist sensitive conversation content, credentials, internal decisions, or personal data without clear informed consent.

Exfiltration Commands

High
Category
Prompt Injection
Content
**File access**: The `-f` flag on `brv curate` reads files from the current project directory only. Paths outside the project root are rejected. Maximum 5 files per command, text and document formats only.

**LLM usage**: `brv query` and `brv curate` send context to a configured LLM provider for processing. The LLM sees the query or curate text and any included file contents. No data is sent to ByteRover servers unless you explicitly run `brv vc push`.

**Cloud sync**: `brv vc push` and `brv vc pull` require authentication (`brv login`) and sync knowledge with ByteRover's cloud service via git. All other commands operate without ByteRover authentication.
Confidence
88% confidence
Finding
send context to

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.