Back to skill

Security audit

code-crawler

Security checks for vulnerabilities and agentic risk

Overview

The scanners rate this skill suspicious because its setup uses unpinned npm packages, and one finding says the setup behavior is not fully disclosed; they did not establish malicious behavior.

Consider postponing installation until the package versions are pinned to reviewed releases and the Cursor setup behavior is clearly disclosed. If you proceed, review the commands and generated configuration before running or approving them. The scanners did not establish malicious behavior, but coverage was partial because several referenced paths were unresolved.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (8)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied chunk performs MCP setup, not graph querying or question answering. Although configuring the same server could support the declared workflow, setup is the entirety of the supplied behavior and targets Cursor's MCP configuration, an operational purpose not disclosed in the description.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The documented npx skills command executes an npm package without specifying a version, and the skill source also lacks an immutable revision. Although this is a skill installer rather than the MCP server itself, a compromised future release could execute code with the user's permissions; no compromise is demonstrated here.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx add-mcp'.

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The registration command leaves both add-mcp and the target MCP package unversioned. Executing an unexpected or compromised release could run code locally or register an unsafe server; the explicit requirement to ask before editing configuration reduces unauthorized changes but does not address package provenance.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx add-mcp'.

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The Cursor setup example runs an unpinned npm configuration utility and names an unpinned server package. This creates a supply-chain exposure during setup and potentially during subsequent server launches, although the project-scoped configuration and consent guidance limit its scope.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx add-mcp'.

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

Even the informational npx add-mcp list-agents command executes package code rather than merely reading static documentation. Without a version pin, a compromised release could execute arbitrary code despite the apparently read-only purpose of listing supported clients.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @nithin554/code-crawler'.

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

The crawl command executes an unpinned version of @nithin554/code-crawler, with -y suppressing the package installation confirmation. Since the tool is intended to read repository contents, a compromised release could access proprietary code or other files available to the process; the supplied document does not establish malicious behavior.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 189)May include surrounding context.

md
call is not what the annotation promises when either is set: `inert: true` means
  self-invocation bypassed the proxy, `proxyBlocked` (`private`, `static`, `final`,
  `final-type`) means the container could never have honoured it. Do not report an
  annotated method as transactional or asynchronous without checking these.
- An edge with `alternatives` is an unresolved *choice*, not an unresolved call: the call site
  is proven but several overloads matched equally, and the target is the first declaration with
  the rivals listed. Confidence is capped at `0.6`. Report both rather than asserting the one

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 228)May include surrounding context.

md
## Do not

- Crawl a monorepo to answer a question about one file.
- Report "no callers" or "dead code" without checking `confidence` (where present) and
  `external` on the edges you did get, and without noting whether dispatch edges were
  followed.
- Treat `find_unresolved` as a bug list. It is a coverage report; some entries are