ClawHub
Back to skill

Security audit

code_flow_graph

Security checks across malware telemetry and agentic risk

Overview

This skill has a coherent code-visualization purpose, but it also runs generated or project data as JavaScript during validation and keeps persistent project-state metadata.

Install only if you are comfortable with a skill reading the target project, writing diagram files, and storing local session metadata. Use caution with existing or untrusted code_flow_graph_data.js files because the validator executes that file as JavaScript; prefer reviewing or replacing the validator before using it on untrusted repositories.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Context-Inappropriate Capability

High
Confidence
98% confidence
Finding
The validator reads an arbitrary user-supplied file and executes it with vm.runInNewContext purely to check syntax and extract globals. That means a crafted data file can run attacker-controlled JavaScript during validation, enabling denial of service via infinite loops or memory exhaustion and potentially broader sandbox-escape risk depending on Node/vm behavior and surrounding environment.

Vague Triggers

Medium
Confidence
82% confidence
Finding
The README instructs the agent to trigger on broad user requests like 'visualize the code architecture of this project' and says an overview should be generated immediately with no scoping questions. In an agent-skill context, vague activation boundaries can cause the skill to run on repositories or modules the user did not clearly intend, increasing the chance of unnecessary codebase analysis, over-collection of context, and surprising side effects such as creating output files or persisting state.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The skill writes files into the user's project directory without disclosing that behavior in the user-facing description. Unexpected filesystem writes can surprise users, alter repositories, and leak intent into version control or shared workspaces, especially when activation is automatic or ambiguous.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill creates a persistent session file containing absolute project paths, timestamps, and analysis history without a privacy or persistence warning. Persistent metadata can expose sensitive project structure and user activity, and may remain on disk longer than users expect or be committed accidentally.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.