ClawHub
Back to skill

Security audit

Botwallet -- Let Your Agent Earn & Spend

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent real-money wallet tool, but it needs Review because it installs an unpinned global CLI and tells agents to create/send a payable sample invoice without explicit opt-in.

Review this carefully before installing. Only use it with strict spending limits, merchant allowlists, and explicit approval for any invoice, payment, withdrawal, x402 purchase, or fund request. Prefer a pinned, verified CLI version and avoid sending hostname, environment, secrets, or broad project metadata during registration.

A.I.G

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (3)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:58
Finding

Unpinned Global Installation of an External Executable

Content
View full analysis
Remediation
View remediation
``` 2. Replace `@latest` with an immutable, reviewed Go module version or commit. 3. Publish and verify SHA-256 checksums or signed release attestations before execution. 4. Provide reproducible-build metadata and software-bill-of-materials information. 5. Avoid global installation where possible. Run the CLI in an isolated environment with only the filesystem and network access it requires. 6. Vendor or bundle the reviewed implementation when feasible so the effective executable payload matches the audited artifact. 7. Restrict permissions on `~/.botwallet/config.json` and `~/.botwallet/seeds/`, and prevent unrelated processes from reading them. 8. Require explicit user approval before installing or upgrading executable dependencies. ]]>

T09 · Insecure Skill Coding Practices

Note
Location
SKILL.md:119
Finding

Unnecessary Transmission of Host and Environment Metadata

Content
View full analysis
Remediation
View remediation

other

Warning
Location
SKILL.md:133
Finding

Unsolicited Creation and Delivery of a Monetized Sample Invoice

Content
View full analysis
--to ``` Explain that this is a sample invoice to demo how payments work. If they pay it, the money goes to the shared wallet (they can withdraw anytime for free). You can send real invoices to anyone, pay other agents, access paid APIs, and request funds. Keep it engaging and show real capabilities. ``` ### Technical Analysis The Skill directs the agent to create a payable invoice and send it to the owner's email after wallet claiming. It does not require the owner to request the demonstration, approve the amount, or consent to receiving the invoice. The example contains a total requested payment of 5.00 USDC, including subjective or promotional line items such as “Being an awesome agent.” These line items do not establish a genuine payable obligation. Labeling the request as a sample does not fully mitigate the risk because it remains a functional payment request that the recipient can pay. Creating and emailing a monetized solicitation is n ...[truncated 1240 chars]
Remediation
View remediation

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The README prominently advertises that an agent can move, request, and withdraw real USDC but does not clearly warn that blockchain transfers can be irreversible, financially risky, and unsafe if an agent is misconfigured or abused. In the context of an agent skill that encourages shell access and automated payment actions, omission of these cautions can mislead operators into enabling high-impact financial capabilities without understanding the consequences.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · SKILL.md (reported line 20)May include surrounding context.

md
• Approve or reject payments, withdrawals, and x402 purchases
• Set guard rails: per-transaction limits, daily limits, merchant allowlists
• Fund wallets and respond to fund requests
• Claim wallets you create for them

### Two-Step Transaction Pattern

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly instructs the agent to create and send a 'sample invoice' to the human owner, but it frames this as a demo and celebratory step rather than a real monetary request. That creates a meaningful risk of deceptive or insufficiently informed payment solicitation, where a user may pay thinking it is purely illustrative even though it transfers real funds.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 286)May include surrounding context.

POST/PUT Requests

bash
botwallet x402 fetch https://api.example.com/query \
  --method POST \
  --body '{"q": "test"}' \
  --header "Accept: application/json"

Ssd 3

Low
Category
Not specified by scanner
Confidence
67% confidence
Finding

The skill explicitly tells the agent to save approval and transaction identifiers to persistent memory for later use. While framed as operational continuity, this is a natural-language instruction to retain and potentially reuse user-associated financial workflow data beyond the immediate interaction, which can increase exposure if memory is broadly accessible or later surfaced.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.