Back to skill

Security audit

Prooflane MCP Skill

Security checks for vulnerabilities and agentic risk

Overview

The skill’s purpose is coherent, but its install path can run mutable external code and an unpinned future package command.

Review this skill before installing. Use a disposable or least-privilege environment, inspect the cloned repository scripts before running them, pin the repository to a trusted commit, and avoid the unpinned npx package form unless a specific reviewed version is available.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Findings (1)

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:38
Finding

Unpinned Remote Repository Is Downloaded and Executed

Content
View full analysis

Vulnerability Details

File Locations:

  • SKILL.md:38-46
  • manifest.yaml:27-32
  • references/INSTALL.md:3-8

Vulnerability Type: Remote payload retrieval and execution from a mutable, unpinned source
Risk Level: High

Vulnerable code in SKILL.md:38-46:

bash
git clone https://github.com/xiaojiou176-open/ui-automation-control-plane.git
cd ui-automation-control-plane
bash
./scripts/setup.sh

Vulnerable code in manifest.yaml:27-32:

yaml
installation:
  repo: https://github.com/xiaojiou176-open/ui-automation-control-plane
  steps:
    - git clone https://github.com/xiaojiou176-open/ui-automation-control-plane.git
    - cd ui-automation-control-plane
    - ./scripts/setup.sh

Vulnerable code in references/INSTALL.md:3-8:

bash
git clone https://github.com/xiaojiou176-open/ui-automation-control-plane.git
cd ui-automation-control-plane
./scripts/setup.sh

Technical Analysis

The installation instructions clone the current default branch of an external GitHub repository and immediately execute its scripts/setup.sh file. No immutable commit hash, signed release, checksum, or other integrity verification is specified.

Consequently, the effective executable payload can change after this skill packet has been reviewed. The remote repository and setup script are not included in the audited artifact, so the behavior of that script cannot be established from the available project files. A compromise of the repository, its maintainer account, or its development workflow could replace the setup script or files invoked by it with attacker-controlled code.

The documented alternative, pnpm install, also creates supply-chain exposure because dependency installation may run package lifecycle scripts. The instructions do not require a reviewed frozen lockfile or suppression of dependency scripts.

Attack Path

  1. An attacker compromises t ...[truncated 1522 chars]
Remediation
View remediation

Remediation Suggestions

  1. Pin the repository to a reviewed, immutable commit rather than executing the current default branch:

    bash
    git clone https://github.com/xiaojiou176-open/ui-automation-control-plane.git
    cd ui-automation-control-plane
    git checkout --detach REVIEWED_COMMIT_SHA
    
  2. Verify the checked-out commit against a trusted, documented value before running any script:

    bash
    test "$(git rev-parse HEAD)" = "EXPECTED_FULL_COMMIT_SHA" || exit 1
    
  3. Prefer a cryptographically signed release or commit and document the exact signature-verification procedure.

  4. Publish and verify a checksum for any release archive or vendored setup script. Obtain the checksum through an independently trusted channel.

  5. Vendor the reviewed setup script into the skill package when practical so its exact contents are covered by the skill audit.

  6. Require users to inspect the setup script and all commands it invokes before execution. Avoid describing remote script execution as a routine trusted step.

  7. Require a committed, reviewed lockfile and use a frozen installation mode, such as:

    bash
    pnpm install --frozen-lockfile
    
  8. Disable package lifecycle scripts where they are unnecessary:

    bash
    pnpm install --frozen-lockfile --ignore-scripts
    

    If lifecycle scripts are required, explicitly enumerate and review the packages permitted to execute them.

  9. Run installation in an isolated, least-privilege environment without production credentials, SSH agents, cloud tokens, or unrelated writable host directories.

  10. Update all duplicated installation instructions in SKILL.md, manifest.yaml, and references/INSTALL.md so they consistently enforce the pinned and verified installation process.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill instructs users to run repository setup and startup scripts (./scripts/setup.sh, ./scripts/dev-up.sh) without any warning that these commands execute arbitrary code from the cloned repository and may modify the local system, install dependencies, or start services. In a security-sensitive agent context, normalizing direct execution of shell scripts from an external repo increases the chance of unsafe code execution and unintended system changes.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The skill recommends executing an unpinned package via npx -y @uiq/mcp-server, which would fetch whatever version is current at execution time once published. That creates a supply-chain risk: users may unknowingly run a compromised, typosquatted, or breaking release, and the skill context explicitly frames this as a command to execute rather than a purely informational reference.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The manifest defines execution via npx -y @uiq/mcp-server without pinning an exact package version. If this path is ever used, it can fetch and run whatever version is current on the registry at execution time, creating a supply-chain risk where a compromised latest release, dependency, or account takeover could result in arbitrary code execution on the local machine.

Content

No source excerpt is available for this finding.