Corporate Credit Memo

• Outcome: Pass
• Latest audit: Sep 11, 2026 · 6:04 AM UTC
• Version: 1.0.0

Overview

This is a coherent credit-memo drafting skill that handles sensitive financial information, but its access and outputs match its stated banking purpose.

Install only if you are authorized to process the borrower's financial statements, ownership details, and deal terms in your ClawHub environment. Treat generated memos as confidential internal drafts, verify web-sourced facts and calculations, and ensure all internal risk, legal, RAROC, RWA, PD/LGD, and policy inputs are completed by qualified bank teams before any credit decision.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Memory Poisoning: Persistent Context Injection, Context Window Stuffing, Memory Manipulation
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (7)

Memory Manipulation

High

• Category: Memory Poisoning

Content

| Industry Policy Compliance | [Yes / Conditional / No] | See Section 1.3 |

### 8.2 Recommendation
Clear statement: Approve / Approve subject to conditions / Decline.

If conditional approval:
| No. | Condition Precedent / Special Requirement | Deadline |

• Confidence: 80% confidence
• Finding: Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: This markdown file describes a workflow where users upload annual reports and deal parameters, then receive a generated credit memorandum. Because credit underwriting materials can contain confidential financial and customer information, the skill description should include an explicit user-facing warning about data sensitivity, retention, and appropriate use to satisfy the markdown-file warning requirement.

Natural-Language Policy Violations

Medium

• Confidence: 95% confidence
• Finding: This is a natural-language policy issue because the README imposes a specific output language by default. The documentation does not indicate user opt-in, configurable language selection, or a clearly justified region-specific constraint, so it appears to force a locale choice.

Natural-Language Policy Violations

Medium

• Confidence: 98% confidence
• Finding: The description states that the skill generates credit memoranda "in English," and the drafting principles later require "formal institutional English." This imposes a language choice by default without offering the user an option to select another language or confirming that English is desired.

Vague Triggers

Medium

• Confidence: 92% confidence
• Finding: The manifest says to use the skill whenever a user wants a broad range of outputs, including generic requests like "assess this borrower" and "analyse this company for a loan." It does not provide exclusion conditions or clear boundaries for when a request should not invoke this skill, increasing the risk of unintended activation for everyday analyst support queries.

Natural-Language Policy Violations

Medium

• Confidence: 96% confidence
• Finding: The instruction to "Write in formal institutional English" is a hard requirement embedded in the workflow. Because no opt-in or alternative language path is provided elsewhere, this conflicts with the language/locale policy requirement.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: This template explicitly solicits highly sensitive ownership and ultimate beneficial owner information, including control arrangements and pledged shares, but provides no guidance on confidentiality, lawful basis, minimisation, or safe handling. In a banking workflow, this can cause the system or end users to collect and reproduce regulated personal and corporate data into generated outputs without appropriate warnings or controls, increasing privacy, compliance, and data leakage risk.

Static analysis

No suspicious patterns detected.