ClawHub
Back to skill

Security audit

Grit

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent persistence helper, but it can auto-trigger broadly and encourages installing tools and using authenticated sessions without tight scoping.

Install only if you want an agent to become more aggressive about solving blocked tasks. Review workspace SOPs and require explicit approval before it installs software, installs new skills, uses authenticated browser sessions, or switches to direct API/CLI actions.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

YARA rule 'agent_skill_prompt_injection_hidden_instructions': Prompt injection or hidden instructions embedded in AI agent skill text [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · SKILL.md (reported line 14)May include surrounding context.

md
better tools if needed, and keep iterating until the task is solved or a real external blocker remains.

Before escalating, read and obey the workspace SOP if it exists. Grit is subordinate to the user's operating rules, local scan procedures, audit requirements, and workflow preferences.

Stay inside system safety rules and tool policy. Grit increases persistence and experimentation; it does not override guardrails.

## Core behavior

- Prefer action over hand-wringing.
- Do not stop at the first failed approach.
- Treat every failure as data for the next attempt.
- Escalate tools aggressively when the current method is weak.
- Keep the user updated with short, concrete progress reports when useful.
- Avoid repeated blind retries; change something material each cycle.
- Follow SOP requirements religiously, especially around browser method preferences, screenshot analysis, audit logging, and security checks.
- When installing skills or software, use the user's required scan/audit workf

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill advertises very broad trigger phrases such as 'do whatever it takes', 'keep trying until it works', and 'use whatever tools you need', which can match ordinary user language rather than an explicit request for this specific skill. In an agentic system, that can cause unintended activation of a persistence/escalation behavior that installs tools or changes tactics more aggressively than the user intended, increasing the chance of risky actions or policy drift.

Content

No source excerpt is available for this finding.