ClawHub
Back to skill

Security audit

Lexi

Security checks for vulnerabilities and agentic risk

Overview

Lexi is a disclosed filesystem audit and cleanup skill with broad but purpose-aligned scanning and user-confirmed changes.

Before installing, users should be comfortable with a broad filesystem audit tool. Use targeted scope for sensitive environments, review the exclusions carefully, and only approve move or archive batches after checking the proposed paths and reference updates.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (6)

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · SKILL.md (reported line 30)May include surrounding context.

md
| `**/.git/` | Git internals — never modify |
| `**/.env` | Environment secrets |
| `**/auth-profiles.json` | Agent auth credentials |
| `**/credentials.json` | Service credentials |
| `**/node_modules/` | Package dependencies — managed by npm/yarn |
| `**/__pycache__/` | Python bytecode cache — auto-generated |
| `**/.venv/lib/` | Virtual environment packages — managed by pip |

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scanning-framework.md (reported line 20)May include surrounding context.

md
| `**/.git/` | Git internals — never modify |
| `**/.env` | Environment secrets |
| `**/auth-profiles.json` | Agent auth credentials |
| `**/credentials.json` | Service credentials |
| `**/node_modules/` | Package dependencies — managed by npm/yarn |
| `**/__pycache__/` | Python bytecode cache — auto-generated |
| `**/.venv/lib/` | Virtual environment packages — managed by pip |

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill advertises very broad natural-language triggers like 'organize', 'clean up', and similar variations for a capability that can eventually move, archive, and update references across a user's filesystem. This raises the chance of accidental invocation from ambiguous user requests, causing unintended large-scale scanning or leading the conversation toward destructive file operations.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
85% confidence
Finding

Reading crontab -l and mining PM2 or agent configs persists visibility into scheduled tasks and operational automation outside the immediate directory tree. Even if intended for dependency mapping, this broadens collection into system/user runtime configuration and may expose sensitive command paths, schedules, or embedded arguments without clear necessity or scoped consent.

Content

Scanner excerpt · SKILL.md (reported line 67)May include surrounding context.

md
4. **Reference scan (critical for safe reorganization):**
   - Grep all `.md` files for path references (absolute and relative)
   - Grep all `.sh`, `.py`, `.js`, `.ts` scripts for hardcoded paths
   - Extract paths from crontab (`crontab -l`)
   - Extract paths from PM2 configs
   - Extract paths from OpenClaw agent configs
   - Map all symlinks with source → target

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The slash-command section again includes ambiguous triggers such as 'organize', 'catalog', and 'clean up my files', which are common phrases that may appear in unrelated conversations. Because this skill has access to broad filesystem auditing and later execution phases, accidental activation materially increases operational risk.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
85% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · scanning-framework.md (reported line 139)May include surrounding context.

md
| **Markdown links** | `.md` files | `[text](path)` or bare paths in prose |
| **Script paths** | `.sh`, `.py`, `.js`, `.ts` | String literals containing `/home/`, `~/`, or relative paths |
| **Config paths** | `.yaml`, `.json`, `.toml` | Path-valued keys |
| **Cron paths** | `crontab -l` | Command strings |
| **PM2 paths** | PM2 ecosystem configs | `script` and `cwd` values |
| **Symlink targets** | Filesystem | `readlink` output |
| **Import statements** | `.py`, `.js`, `.ts` | `import`, `require`, `from` with paths |

Static analysis

No suspicious patterns detected.