Back to skill

Security audit

Projitive

Security checks for vulnerabilities and agentic risk

Overview

The skill is coherent for repository governance, but it asks users to globally install a mutable third-party MCP package and can initialize project governance files automatically.

Review the MCP package before installing, prefer a pinned version or controlled local install, and only use the skill in repositories where automatic `.projitive/` governance files and task updates are acceptable.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:13
Finding

Unpinned Global Installation of a Mutable Third-Party Package

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 13–19
Vulnerability Type: Supply-chain exposure through an unpinned dependency installed globally
Risk Level: Medium

Vulnerable Code:

markdown
Install (always use latest):

```bash
npm install -g @projitive/mcp@latest

@projitive/mcp must be kept up to date — outdated versions may be missing tools or contain bugs.

text

### Technical Analysis

The skill instructs the operator or agent to install the mutable `latest` release of `@projitive/mcp` globally. It does not pin a reviewed version, enforce package integrity, provide a lockfile, or require approval before installation.

Because the MCP implementation is not included in this repository, its effective behavior cannot be determined from the audited files. The contents associated with the `latest` distribution tag can also change after this skill has been reviewed. Depending on npm behavior and package configuration, installation may execute package lifecycle scripts. A global installation increases the potential scope beyond the current project and makes the installed tool available to other repositories and sessions.

No evidence establishes that the current `@projitive/mcp` package is malicious. The vulnerability is the unsafe, non-reproducible dependency acquisition process and the resulting supply-chain exposure.

### Attack Path

1. An attacker compromises the npm package, a maintainer account, publishing credentials, or the upstream release process.
2. The attacker publishes a malicious version and assigns it to the `latest` distribution tag.
3. An agent or operator follows the skill's mandatory installation instruction.
4. npm downloads and globally installs the attacker-controlled release without validation against a previously reviewed version or integrity value.
5. Malicious lifecycle code may execute during installation, or malicious behavior may run when the installed MC
...[truncated 829 chars]
Remediation
View remediation

Remediation Suggestions

  1. Replace @latest with an exact, reviewed package version.
  2. Prefer a project-local dependency recorded in package.json and a committed lockfile instead of a global installation.
  3. Verify package provenance and integrity using registry integrity metadata, trusted publishing, signatures, or an approved internal artifact registry.
  4. Review the selected package version, including lifecycle scripts and transitive dependencies, before approving it.
  5. Disable lifecycle scripts during acquisition where operationally feasible, and run required setup separately after review.
  6. Require explicit operator approval before installing or upgrading external software.
  7. Execute the MCP server in a least-privileged sandbox with narrowly scoped filesystem and network access.
  8. Use a controlled update process that reviews and tests each new version before changing the approved pin.
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 (2)

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
95% confidence
Finding

The instruction Do NOT ask the user to do this manually encourages autonomous action that bypasses a normal approval step for repository-modifying behavior. In the context of a governance skill that can create .projitive/ state and related artifacts, this increases the chance of unauthorized or unexpected modifications and weakens user control over agent actions.

Content

Scanner excerpt · SKILL.md (reported line 29)May include surrounding context.

md
# Init: Repo Governance Setup

When `.projitive/` does not exist or is incomplete, call `projectInit(projectPath="<project-dir>")` immediately. Do NOT ask the user to do this manually.

# Mandatory Governance Files

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly instructs the agent to initialize repository governance state immediately when .projitive/ is missing or incomplete, which can create or modify files without prior user awareness or consent. In an agent setting, automatic repo writes are risky because they may alter project state, introduce unwanted artifacts, or overwrite expected workflows before the user has approved those changes.

Content

No source excerpt is available for this finding.