Back to skill

Security audit

Defi Sniper

Security checks for vulnerabilities and agentic risk

Overview

The skill is transparent about helping execute bounded DeFi trades, but its setup instructions use unpinned latest installers and a broad update-all command in a wallet/trading context.

Review this skill carefully before installing. Use a dedicated, least-privileged environment without wallet secrets during setup, avoid `update --all`, pin reviewed installer and dependency versions where possible, and require manual confirmation for any live trade unless you have separately audited the execution path and limits.

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:22
Finding
Unpinned Third-Party Installation and Unscoped Global Updates## Vulnerability Details **File Location**: `SKILL.md`, lines 22-29 **Vulnerability Type**: Mutable and unpinned third-party dependencies **Risk Level**: Medium The Skill recommends installing dependencies by executing the latest available version of a remote npm package and subsequently updating every installed Skill: ```bash Install/update: ```bash npx -y clawhub@latest install minara npx -y clawhub@latest install torchmarket npx -y clawhub@latest install torchliquidationbot npx -y clawhub@latest update --all ``` ``` ### Technical Analysis The command `npx -y clawhub@latest` retrieves and executes a mutable package release without pinning its version or verifying its integrity. Although the document identifies the inspected upstream Skill versions as `minara 1.1.9`, `torchmarket 4.2.7`, and `torchliquidationbot 3.0.2`, the installation commands do not constrain installations to those reviewed versions. Consequently, the code and instructions installed when a user follows this documentation may differ from the versions that were inspected. The `-y` option suppresses the normal installation confirmation, reducing the opportunity for the user to review the package being downloaded. The `update --all` operation also exceeds the minimum scope required for this Skill. It can modify unrelated installed Skills, potentially introducing unreviewed behavior outside the three declared dependencies and invalidating prior security review results. ### Attack Path 1. An attacker compromises the npm package identified as `clawhub`, a future package release, its publishing account, or one of the remotely distributed Skill packages. 2. The attacker publishes a malicious or unsafe release under the package or Skill name expected by these commands. 3. A user follows the documented installation procedure. 4. `npx -y clawhub@latest` downloads and executes the mutable package without a pinned version or integrity constraint. 5. T ...[truncated 1083 chars]
Remediation
## Remediation Suggestions 1. Pin the `clawhub` npm package to a specifically reviewed version rather than using `@latest`. 2. Pin each installed Skill to the exact reviewed version: `minara 1.1.9`, `torchmarket 4.2.7`, and `torchliquidationbot 3.0.2`, subject to the package manager's supported syntax. 3. Use npm lockfiles, package integrity hashes, signed release metadata, or an approved internal registry where supported. 4. Remove `npx -y` or require explicit operator confirmation before downloading and executing a package. 5. Remove `update --all`. Update only the three named dependencies, one at a time, after reviewing their release contents. 6. Run installation under a dedicated, least-privileged account without access to wallet secrets or production credentials. 7. Separate dependency installation from live operation so credentials such as `MINARA_API_KEY` and signer material are not present during installation. 8. Re-audit dependency updates before approving them for live trading environments.
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 (5)

Ae1

High
Category
analysis-evasion
Content
- Minara inspected docs describe intent parsing and transaction assembly, but do not expose a dedicated "mempool scanner" endpoint in the inspected `SKILL.md`.
Confidence
100% confidence
Finding
Referenced artifact was not completely inspected

Rp1

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding
The skill instructs users to execute `npx -y clawhub@latest install minara`, which pulls and runs the latest published package code at execution time without pinning to a reviewed version. If the upstream package or its dependency chain is compromised, a user invoking this setup command could execute attacker-controlled code on the local machine.

Rp1

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding
This command uses `npx -y clawhub@latest install torchmarket`, which causes dynamic retrieval and execution of the newest package version. In a skill that later handles API keys, RPC endpoints, and potentially trading credentials, executing an unpinned installer meaningfully increases supply-chain risk.

Rp1

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding
The instruction `npx -y clawhub@latest install torchliquidationbot` relies on an unpinned, network-fetched executable package. Because this skill is for DeFi execution orchestration and may be used in environments with wallet access or vault-related configuration, compromise of the fetched package could directly lead to credential theft or malicious transaction preparation.

Rp1

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding
The update instruction `npx -y clawhub@latest update --all` is especially risky because it requests broad updates while also executing the newest unpinned package manager code. That combination can unexpectedly alter multiple installed skills at once, increasing the blast radius of a compromised or buggy upstream release.

Static analysis

No suspicious patterns detected.