Paper Reading Experiment Design

• Outcome: Pass
• Latest audit: Sep 16, 2026 · 9:38 AM UTC
• Version: 0.1.0

Overview

This skill is a structured research workflow that creates local experiment-design notes and includes clear user review gates.

Install this if you want an agent to help structure experiment-design documents from your research notes. Review the workspace root and source artifacts before use, since it will read relevant local research files and write multiple markdown planning files.

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

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Autonomous Decision Making

Medium

• Category: Excessive Agency
• Confidence: 80% 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 54)May include surrounding context.

md

This is a stepwise workflow, not a one-shot artifact generator. Ask one bounded decision packet at a time, provide the recommended answer, wait for the user's confirmation, revision, or explicit delegation, then update the relevant artifact before moving to dependent decisions.

Use concise decision packets. A packet should show only the decision now needed, why it matters, the recommended answer, and the consequence of accepting it. Do not ask the user to approve a whole artifact when one unresolved decision would redirect downstream work.

Do not treat silence, time pressure, or a broad opening instruction such as "you decide" as approval for every later gate. Delegation counts only after the user has seen the specific gate or packet recommendation. Delegation is scoped to that gate or packet and must be recorded in the relevant artifact.