ClawHub
Back to skill

Security audit

Map Is Not the Territory

Security checks for vulnerabilities and agentic risk

Overview

This is a non-executable coaching skill for checking whether metrics, models, or plans still match reality; the scanner flags are ordinary prose, not unsafe instructions.

This skill is reasonable to install as a decision-making aid. Treat it as a coaching framework, not a source of authority; if you record real organizational examples, avoid including confidential or sensitive details unless you intentionally want them stored.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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 · SKILL.md (reported line 73)May include surrounding context.

md
|---|---|---|---|
| Startup growth | MAU | Churn drivers, value realization | Users activate but don't retain |
| Sales | Pipeline × close rate | Buying-committee dynamics | Close rate collapses late-stage |
| Team health | Satisfaction scores | Silent disengagement | Top performers leave without warning |
| Financial model | Revenue/cost projections | Cash flow dynamics | Model shows profit; cash crisis arrives |
| Org chart | Formal authority | Informal influence networks | Decisions route through uncharted nodes |

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
- Team disagreement cannot be resolved by data — parties are disagreeing about maps, not territory
- A strategy, model, process, or org chart has not been reviewed in more than one planning cycle
- "The data shows..." but the data is old, narrow, or from a proxy source
- A metric is being optimized directly without asking whether it still maps to the underlying goal
- An AI model's benchmark score, leaderboard rank, or internal "world model" is being treated as proof of real-world capability (AI adoption / AI hype)

**Not when:** the map is demonstrably current and well-calibrated; decision is low-stakes; territory is stable and map is freshly validated.

Static analysis

No suspicious patterns detected.