Back to skill

Security audit

Genexxis security checks: CVE, CI lint, domain audit, prompt injection

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because it repeatedly runs an unpinned `agentcash@latest` package and sends file contents to an external service without a warning about sensitive data; this does not establish malicious intent.

Consider avoiding installation until the CLI version is fixed and verified, and until you can confirm what data is sent to the external service and redact or withhold sensitive content. The scan is partial: it inspected the listed Markdown file but could not resolve several local references, so the evidence is incomplete.

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
  • 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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
Findings (21)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · SKILL.md (reported line 81)May include surrounding context.

md
`POST https://ciaudit.genexxis.com/dockerfile` · $0.01 · MCP tool `ciaudit_dockerfile`

Lint a Dockerfile: base image pinning, root user, curl|sh, remote ADD, secrets in ENV/ARG, apt/pip/npm hygiene, chmod 777, EXPOSE 22, HEALTHCHECK, layer cache order.

Parameters: `content` (required).

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · SKILL.md (reported line 86)May include surrounding context.

Parameters: content (required).

bash
npx agentcash@latest fetch "https://ciaudit.genexxis.com/dockerfile" -m POST -b '{"content":"FROM node:latest\nRUN curl -sL https://example.com/install.sh | sh\nCOPY . .\nCMD [\"node\", \"server.js\"]\n"}' --max-amount 0.01

Lint a docker-compose file

Docker Socket Access

High
Category
Privilege Escalation
Confidence
90% confidence
Finding

Potential security issue detected. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 98)May include surrounding context.

Parameters: content (required).

bash
npx agentcash@latest fetch "https://ciaudit.genexxis.com/compose" -m POST -b '{"content":"services:\n  web:\n    image: nginx:latest\n    privileged: true\n    volumes:\n      - /var/run/docker.sock:/var/run/docker.sock\n"}' --max-amount 0.01

Grade HTTP security headers (A-F)

Privileged Kubernetes Workload

High
Category
Tool Misuse
Confidence
70% confidence
Finding

Code deploys a privileged Kubernetes workload (privileged container, hostPath mount, or host namespaces). This grants root on the node and is a node/cluster takeover vector.

Content

Scanner excerpt · SKILL.md (reported line 98)May include surrounding context.

Parameters: content (required).

bash
npx agentcash@latest fetch "https://ciaudit.genexxis.com/compose" -m POST -b '{"content":"services:\n  web:\n    image: nginx:latest\n    privileged: true\n    volumes:\n      - /var/run/docker.sock:/var/run/docker.sock\n"}' --max-amount 0.01

Grade HTTP security headers (A-F)

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

Medium
Category
MCP Rug Pull
Confidence
99% confidence
Finding

The CLI invocation uses agentcash@latest, a mutable npm tag rather than an immutable version. Although this is a payment CLI rather than an MCP server, executing a compromised future release could expose local files and wallet credentials accessible to the process.

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 accounts example executes a package selected by the mutable latest tag in a wallet-related context. A compromised release could access available wallet material or misrepresent deposit addresses; this is another occurrence of the same unpinned dependency issue.

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 price-check command also executes agentcash@latest; a free remote operation does not make local package execution safe. This shares the same supply-chain root cause as the other findings on line 22, rather than representing a separate compromise.

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 package-scan example executes agentcash@latest without an immutable version. A compromised npm release could run with the agent's local privileges; the payment cap does not constrain malicious CLI code.

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 CVE lookup example resolves executable code through agentcash@latest. Although the lookup itself is read-only, a compromised CLI release could access local data and wallet 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 KEV lookup runs a CLI selected by a mutable npm tag. This creates conditional supply-chain code-execution risk even though the advertised API operation only retrieves vulnerability information.

Content

No source excerpt is available for this finding.

Remote content checks lack a warning about disclosing potentially sensitive repository data.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The lint endpoints require transmitting workflow, Dockerfile or Compose content to an external service without instructions to redact secrets or obtain authorization for that disclosure. These files can contain literal credentials and private infrastructure details, and approval to incur a small charge does not necessarily authorize sharing repository content.

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-lint example uses agentcash@latest rather than a reviewed fixed release. A compromised release could access repository contents and credentials available to the agent, beyond the intended submitted 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 Dockerfile-lint command executes an unpinned CLI release through agentcash@latest. The supply-chain risk comes from this executable dependency, not from the Dockerfile text supplied as JSON.

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 Compose-lint example runs agentcash@latest, permitting dependency changes without review. A compromised CLI could execute with local privileges independently of the harmless treatment of the Compose sample as request 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 headers-check example uses a mutable npm release tag for executable code. A compromised release could expose local secrets even though the intended request only checks a public URL.

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 TLS-check example executes agentcash@latest without release pinning. The passive nature of the remote check does not mitigate malicious code introduced through a compromised CLI release.

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 email-check example obtains executable CLI code using the mutable latest tag. If the dependency is compromised, local data and wallet access could be abused despite the request's read-only purpose.

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 domain-audit example executes agentcash@latest rather than a fixed reviewed artifact. A malicious future release could run with the agent's permissions; endpoint and payment restrictions would not contain that code.

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 SEO example relies on a mutable package tag for local CLI execution. This introduces supply-chain risk unrelated to whether the inspected website is public or the API request is passive.

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 bundled-audit command executes agentcash@latest, allowing executable code to change without explicit review. A compromised release could access local credentials or funds beyond the documented payment cap.

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 prompt-injection-check example runs an unpinned CLI release. Its security purpose does not establish trust in future package versions, which could execute malicious code with the agent's privileges.

Content

No source excerpt is available for this finding.