ClawHub
Back to skill

Security audit

Cynefin

Security checks for vulnerabilities and agentic risk

Overview

This is a Markdown-only decision-framework skill that does not request system access, credentials, persistence, or unsafe actions.

This skill appears safe to install as a decision-support aid. Users should treat its recommendations as a thinking framework, not as authority to take irreversible business, operational, or crisis actions without normal review.

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
  • 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 (2)

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
80% confidence
Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · examples/snowden-at-ibm-1999-and-the-hbr-synthesis-2007.md (reported line 65)May include surrounding context.

md
**First, the most expensive management error is using a Complicated-domain method on a Complex problem.** Symptoms: detailed plans that fail in execution, post-mortems that blame "execution" rather than method, recurrent failure of expert recommendations, "best practices" that don't work. The fix: switch to Probe-Sense-Respond, run multiple safe-to-fail experiments, amplify what works.

**Second, the boundary between Clear and Chaotic is the most dangerous.** Long success in a Clear domain (routine operations, mature market, established practice) breeds complacency. A small change (new competitor, regulatory shift, technology disruption) can push the situation into Chaotic without warning. Leaders who respond with "do what we always do" preside over the cliff fall. The defense: explicit monitoring for boundary signals; sensors deliberately positioned to catch the transition.

**Third, the framework is diagnostic, not prescriptive.** Cynefin does not tell you what to do; it tells you what *kind* of doing to do. The actual content of each domain's action (which experiments, which experts, which SOP, which decisive crisis intervention) requires domain expertise. Treating Cynefin as a recipe book reduces its value; treating it as a diagnostic lens preserves it.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 18)May include surrounding context.

md
- A familiar approach has stopped working and you can't articulate why
- Experts disagree on the right answer — a crisis unfolding where the previous playbook doesn't apply
- "Best practices from X" imported without checking if the domain matches
- A team is over-planning something emergent, or "let's get more data" when data won't come without action
- Allocating AI capex or racing AI-native competition: deciding which AI bets are engineering (Complicated), emergent agent/adoption experiments (Complex), or live incidents (Chaotic)

Static analysis

No suspicious patterns detected.