ClawHub
Back to skill

Security audit

Lucky Coding Agent

Security checks for vulnerabilities and agentic risk

Overview

The skill is instruction-only and matches its coding-agent purpose, but it encourages autonomous no-sandbox agents that can change code, push branches, create PRs, and signal OpenClaw without clear approval checkpoints.

Install only if you are comfortable letting your assistant spawn local coding-agent CLIs that may edit projects using your existing accounts. Prefer sandboxed or temp worktrees, avoid `--yolo`, set narrow task scope and timeouts, review changes before commits, and require explicit approval before any push, PR, or OpenClaw event command.

SkillSpector

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 (3)

Context-Inappropriate Capability

Medium
Confidence
88% confidence
Finding
The skill instructs pushing branches and creating pull requests, which extends it from local coding delegation into remote publication and repository mutation. Because this can publish agent-generated code externally, it increases the risk of unintended code release, policy bypass, or propagation of unsafe changes without a distinct approval checkpoint.

Context-Inappropriate Capability

Medium
Confidence
91% confidence
Finding
The auto-notify section instructs spawned agents to invoke `openclaw system event`, giving the delegated subprocess an additional platform-signaling capability beyond coding. That creates a cross-boundary action path where an agent can emit events into the host platform, which could be abused for spoofed status notifications or unintended orchestration triggers.

Missing User Warnings

High
Confidence
96% confidence
Finding
This instruction combines `codex --yolo` with 'Commit and push,' enabling unsandboxed autonomous code changes and immediate remote publication. That is dangerous because it can modify the local repository, create commits, and publish them upstream without meaningful review, potentially shipping malicious, broken, or out-of-scope changes.

Static analysis

No suspicious patterns detected.