Security audit

Jarvis Mission Control for OpenCLAW AIS

Security checks across malware telemetry and agentic risk

Overview

This is a documentation-only skill for a self-hosted mission-control dashboard, with sensitive features disclosed but worth reviewing before use.

Before installing or running the linked app, audit the GitHub code and dependencies, keep the dashboard local or protected, use least-privilege GitHub and MissionDeck credentials, avoid exposing the SOUL editor to untrusted users, and back up .mission-control before using reset commands.

SkillSpector

By NVIDIA

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 (5)

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The skill advertises Claude Code session tracking that auto-discovers JSONL data under ~/.claude/projects/, but the description does not clearly warn users that local session history and metadata from their home directory may be read and surfaced in the dashboard. This creates a meaningful privacy and confidentiality risk because users may install or run it without understanding that local agent activity, branches, token usage, and related session details are being collected.

Missing User Warnings

Medium

Confidence: 87% confidence
Finding: The GitHub Issues Sync feature states that users should configure GITHUB_TOKEN and GITHUB_REPO, but it does not explicitly warn that a personal access token is used to access external GitHub services and that repository data will be transmitted off-host. Users may underestimate the sensitivity of the token and repo contents, increasing the chance of over-privileged tokens, accidental disclosure, or unintended external data sharing.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The browser-based SOUL editor is described as allowing direct viewing and editing of SOUL.md, MEMORY.md, and IDENTITY.md files, but the skill text does not prominently warn that this can modify agent files on disk from a web interface. That omission is dangerous because it lowers user awareness of a high-impact capability that could alter agent behavior, persist tampering, or corrupt important local state if the dashboard is exposed or misused.

Missing User Warnings

Low

Confidence: 78% confidence
Finding: The Telegram auto-routing feature can create task cards automatically when a message mentions an agent bot, but the description does not clearly warn that this happens without manual review. This can enable spam, social engineering, or unwanted task injection into operational workflows if the integration is enabled without users understanding the automation behavior.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The document includes destructive reset commands that permanently delete task data and the entire .mission-control directory, but it does not provide any warning, confirmation step, backup guidance, or scope-limiting safeguards. In an agent skill context, such commands may be copied or automated by users or agents, increasing the chance of accidental irreversible data loss.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.