Autonomous Project Management
Security checks across malware telemetry and agentic risk
Overview
This skill is a plausible project-management helper, but it asks agents to spawn autonomous subagents, modify files, run an unbundled script, and commit to git without clear user approval or scope limits.
Install only in a controlled workspace. Replace the hardcoded local path, inspect or supply the referenced project-manager.cjs script before use, and require the agent to ask before spawning subagents, modifying project files, or running git add/commit. Avoid using it in repositories with secrets or sensitive internal state unless STATE.yaml contents and commit scope are tightly limited.
SkillSpector
By NVIDIA
Vulnerability Patterns
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
VirusTotal
No VirusTotal findings