Back to skill

Security audit

base-gas

Security checks for vulnerabilities and agentic risk

Overview

The skill is rated suspicious with medium confidence because scanners report unpinned, changeable package installs and transaction advice that may proceed using unvalidated fees.

Consider the verdict a reason to review carefully before installing. The reported risks may be reduced by using a reviewed, fixed package version and ensuring transaction fees are checked against limits and signing follows your authorization. Skillspector completed scans of the three listed components but marked its analysis partial because some local references could not be resolved; the AIG scan completed with no findings. This evidence does not establish that the skill or any package release is safe.

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
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (18)

The transaction workflow instructs the agent to sign and proceed without warning the user about transaction risks.

High
Category
Output Handling
Confidence
94% confidence
Finding

The workflow tells the agent to sign using API-supplied fees directly and to proceed based on the service's advice, without specifying independent fee bounds or an authorization check. A compromised service could supply excessive priority fees and cause financial loss; the concrete issue is unvalidated financial input, not merely the absence of a generic warning.

Content

No source excerpt is available for this finding.

Manifest references MCP server without version pin: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The manifest instructs agents to use agentcash@latest, exposing them to mutable executable code from npm. This duplicates the line-23 supply-chain issue rather than identifying a separate MCP server, but the wallet-enabled execution risk is real.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The instruction runs agentcash@latest, a mutable npm release, rather than a reviewed, fixed version. A compromised release could execute with the agent's permissions and potentially access wallet credentials or funds; this is a CLI supply-chain risk even though the finding labels it an MCP server.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

This gas-query example executes the mutable agentcash@latest package. An upstream compromise could turn an ordinary price lookup into local code execution with access to the agent's environment and potentially its payment wallet.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The parameterized gas-query example still executes agentcash@latest without fixing the executable version. A malicious future release could access local secrets or wallet capabilities rather than merely fetching gas data.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The advice request invokes a mutable npm release using agentcash@latest. Compromise of that release could execute attacker-controlled code in a workflow associated with transaction signing and wallet access.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The chain-comparison command executes agentcash@latest, whose code can change without changes to this skill. A compromised package could execute with the agent's privileges and expose secrets or payment credentials.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The history lookup depends on executing the unpinned agentcash@latest package. An upstream supply-chain compromise could therefore affect the local agent and its wallet, not just the accuracy of historical data.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The cheapest-window request uses agentcash@latest, allowing the executable dependency to change over time. A compromised release could run arbitrary code with the permissions and credentials available to the payment client.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

This transaction-workflow example executes the mutable agentcash@latest release. A compromised dependency could gain local execution in a particularly sensitive context where transaction and wallet capabilities may be available.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The workflow's history command executes agentcash@latest rather than a fixed, reviewed release. An upstream compromise could expose the agent's local environment or wallet capabilities even though the intended operation is only a data query.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The operation-pricing example invokes the mutable agentcash@latest package. A malicious release could execute under the agent's account and potentially compromise credentials or funds used by the paid-query workflow.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The chain-selection workflow executes agentcash@latest, so its executable code is not fixed by the skill. Compromise of a future npm release could give an attacker the client's local privileges and access to available wallet capabilities.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The scheduling example runs the mutable agentcash@latest package. If incorporated into recurring automation, a compromised release could execute during a later run with whatever secrets and wallet permissions the job retains.

Content

No source excerpt is available for this finding.

The workflow instructs the agent to sign a transaction based on API advice without a warning about transaction authorization or financial consequences.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

This source explicitly instructs the agent to sign with remotely supplied suggestedFees directly when the API returns send_now, without specifying fee validation or spending bounds. Although the text assumes an impending transaction rather than explicitly bypassing authorization, a compromised API could supply excessive priority fees and cause avoidable financial loss.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

Although this is a CLI invocation rather than an MCP server, npx agentcash@latest balance uses a mutable version tag and can execute an unreviewed package release. A compromised release could run with the agent's permissions and potentially access wallet credentials or other local secrets, making the wallet-management context particularly sensitive.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

npx agentcash@latest redeem YOUR_CODE executes a package selected through a mutable tag rather than an exact reviewed release. If a future release is compromised, it could capture the supplied redemption code and potentially access wallet credentials or misuse the CLI's wallet authority; the provided content does not demonstrate such malicious behavior.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

npx agentcash@latest accounts can execute a newly published, unreviewed version of the wallet CLI. A compromised release could substitute attacker-controlled deposit addresses or links, or access locally available wallet secrets, creating a direct financial risk in this funding workflow.

Content

No source excerpt is available for this finding.