Back to skill

Security audit

Conviction FM

Security checks for vulnerabilities and agentic risk

Overview

The skill is coherent for a crypto prediction competition, but it asks users to run an unpinned npm MCP server and describes automatic strategy execution without enough install or action scoping.

Review before installing. Only use this if you intend to use conviction.fm, pin and verify the conviction-mcp package where possible, run it with limited local permissions, and confirm agent creation, position entry, and automatic execution settings before allowing it to act.

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:47
Finding

Unpinned Third-Party MCP Package Is Downloaded and Executed Automatically

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 47–52
Vulnerability Type: Supply-chain risk caused by unpinned remote package execution
Risk Level: Medium

Vulnerable Code

json
{
  "conviction": {
    "command": "npx",
    "args": ["-y", "conviction-mcp"]
  }
}

Technical Analysis

The MCP configuration invokes npx -y conviction-mcp without pinning the package to an exact, reviewed version or verifying its integrity. If the package is not already available locally, npx can retrieve it from the configured npm registry and execute its package-defined code. The -y option suppresses the installation confirmation.

Consequently, the code that runs can change after this Skill has been reviewed. Security depends on the continuing integrity of the npm account, package releases, package dependencies, registry resolution, and the user's npm configuration. The repository contains no lockfile, integrity hash, provenance requirement, or sandbox policy that constrains this execution.

This finding identifies an unsafe dependency-execution pattern; the audited file does not establish that the current conviction-mcp package is malicious.

Attack Path

  1. An attacker compromises the conviction-mcp publishing account, one of its transitive dependencies, or the registry/package-resolution path.
  2. The attacker publishes or causes resolution to a malicious package version.
  3. A user applies the documented MCP configuration.
  4. npx -y conviction-mcp retrieves the currently resolved package without requesting confirmation.
  5. The package's executable code runs under the operating-system account and environment of the Agent process.
  6. Malicious code can attempt to access resources available to that account and process, including readable files, environment variables, network services, and MCP-related data.

Impact Assessment

Successful exploitation could provide arbitrary code ...[truncated 536 chars]

Remediation
View remediation

Remediation Suggestions

  1. Pin conviction-mcp to an exact, reviewed version rather than relying on the latest registry resolution.
  2. Verify package ownership, source repository, release provenance, and npm registry metadata before use.
  3. Use lockfile and integrity verification where supported, and review both direct and transitive dependencies.
  4. Remove -y so installation or retrieval requires explicit operator confirmation.
  5. Prefer a locally installed, audited artifact or a package obtained through a controlled internal registry.
  6. Run the MCP server in a sandbox or container with a read-only filesystem where practical, restricted environment variables, minimal network access, and no unnecessary host credentials.
  7. Ensure wallet and Agent permissions follow least privilege, even if the documented currency is limited to Solana devnet test assets.
  8. Establish a controlled update process requiring security review before changing the pinned package version.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (1)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The activation guidance is overly broad: 'use this skill when the user wants to compete in crypto price prediction pools, create a trading strategy agent, check the leaderboard, or interact with conviction.fm' can match general trading or crypto-assistant requests without clearly requiring explicit user intent to use this specific platform. In an agent ecosystem, this can cause the skill to activate inappropriately and steer users into automated trading-like actions or account/tool operations they did not specifically request.

Content

No source excerpt is available for this finding.