ClawHub
Back to skill

Security audit

Jane Street Puzzle Archivist

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent Jane Street puzzle workflow helper, with a disclosed but review-worthy instruction to update its own knowledge when reusable techniques are learned.

Before installing, treat this as a repository-maintenance skill: let it modify puzzle archive files only in the intended repo, and review any changes to SKILL.md or published versions before accepting them.

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
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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 (1)

Self-Modification

High
Category
Rogue Agent
Confidence
95% confidence
Finding

The skill explicitly instructs the agent to 'update this skill,' which is a self-modification capability. Even though the stated purpose is benign knowledge capture, self-editing instructions are risky because an LLM can persist prompt injection, unsafe workflow changes, or incorrect operational guidance into future runs, turning a one-time compromise or mistake into a durable one.

Content

Scanner excerpt · SKILL.md (reported line 27)May include surrounding context.

md
- `references/solving-patterns.md`
5. Solve the puzzle with a reproducible script stored in the month folder.
6. Record the answer and submission status in that month's `README.md` and `submission.json`.
7. If you learned a reusable technique, update this skill or the repo knowledge files before finishing.

## Required structure