ClawHub
Back to skill

Security audit

Tulebank

Security checks for vulnerabilities and agentic risk

Overview

This skill is coherent and not malicious, but it gives an agent high-impact authority to move and swap funds with only conversational confirmation.

Review this before installing because it can initiate real financial transfers and swaps. Only use it if you trust the tulebank CLI and are comfortable giving an agent access to wallet, KYC, beneficiary, and transaction workflows; for fund movement, prefer exact destination identifiers, manual send mode, and a fresh confirmation of amount, recipient, and token before execution.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (3)

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
91% confidence
Finding

This skill enables irreversible financial actions to be executed by the agent through an auto-send flow, and it explicitly supports bypassing the CLI confirmation with --yes. Although the skill says to obtain human confirmation first, the presence of a one-shot autonomous transfer path increases the risk of prompt-injection, misunderstanding, beneficiary resolution errors, or social-engineering-induced unauthorized transfers.

Content

Scanner excerpt · SKILL.md (reported line 106)May include surrounding context.

text
tulebank send --to <CVU_OR_ALIAS> --amount <n> --token <USDC|wARS>

Creates off-ramp session and automatically sends tokens from the configured wallet to the Ripio deposit address on Base. Shows a confirmation prompt before sending. Add --yes to skip the prompt (only after human has confirmed to you). Use --manual to skip auto-send entirely.

Swap tokens on Base

text

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% confidence
Finding

The skill allows the agent to perform token swaps and skip the CLI confirmation with --yes, which can directly alter asset holdings and expose the user to slippage, price movement, and unintended conversion risk. In a financial skill, letting an LLM initiate trades after only conversational confirmation is dangerous because ambiguity or prompt manipulation can lead to irreversible transactions.

Content

Scanner excerpt · SKILL.md (reported line 112)May include surrounding context.

text
tulebank swap --from USDC --to wARS --amount 10

Swaps tokens via DEX on Base. Default slippage is 100 bps (1%). Use --slippage <bps> to change. Add --yes to skip confirmation (only after human has confirmed to you).

Check off-ramp session status

text

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

Multiple example prompts direct the agent to interact with the user in Spanish, such as "le mando" and "lo agrego y le envío," but the skill does not state that Spanish is optional or limited to a justified locale-specific workflow. This can violate language/locale policy because it implicitly forces one language for user-facing interactions.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.