Back to skill

Security audit

Agent Audit Shield

Security checks for vulnerabilities and agentic risk

Overview

This skill presents itself as a local command-security interceptor, but the reviewed artifact mostly points users to run an unpinned external npx command instead of providing auditable safeguards in the package.

Review this carefully before installing. The artifact itself does not show malicious instructions, but following its usage command may execute whatever `openclaw` package version npx resolves at that time. Prefer an exact pinned version, verified publisher/source, integrity or lockfile checks, and a sandboxed run with minimal filesystem and environment access.

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:20
Finding
Unpinned External Package Execution Through npx## Vulnerability Details **File Location**: `SKILL.md`, lines 20-23 **Vulnerability Type**: Unpinned and unverifiable third-party dependency execution **Risk Level**: Medium **Vulnerable Code Snippet**: ```markdown ## Usage ```bash npx openclaw skill run agent-audit-shield --hardened ``` ``` ### Technical Analysis The documented usage command executes `openclaw` through `npx` without specifying an exact package version, integrity hash, lockfile, or verified source. Depending on the local environment and package cache, `npx` may download executable package content from an external package registry at invocation time. The audited project contains only `SKILL.md`; it does not include the interceptor, scanner, approval interface, or safeguards described in the documentation. Consequently, the effective implementation is external to the reviewed artifact and can differ from the code that was originally assessed. A compromised upstream release, package-name takeover, registry compromise, or malicious future update could therefore cause arbitrary package code to run when a user follows the documented command. This is primarily an insecure software supply-chain dependency rather than evidence that the current upstream package is malicious. ### Attack Path 1. An attacker compromises the package publisher, registry account, distribution channel, or another externally resolved component. 2. The attacker publishes a malicious version under the package name resolved by `npx`, or alters externally resolved skill content. 3. A user follows the documented usage command without an explicit version or integrity constraint. 4. `npx` retrieves or selects the attacker-controlled package version. 5. Package lifecycle logic or the invoked command executes with the permissions of the user running `npx`. 6. The malicious code can access resources available to that user, including local files, environment variables, credentials exposed to ...[truncated 564 chars]
Remediation
## Remediation Suggestions 1. Pin `openclaw` to an exact, reviewed version rather than relying on an unqualified package name. 2. Maintain and distribute a lockfile containing registry-resolved integrity metadata. 3. Document the authoritative package registry and source repository so users can verify provenance. 4. Verify package signatures or cryptographic hashes before execution where supported. 5. Disable or carefully review package lifecycle scripts and transitive dependencies. 6. Bundle the actual skill implementation in the audited artifact, or provide immutable references to all externally executed components. 7. Run the package in a sandbox or container with minimal filesystem access, restricted environment variables, limited network access, and no elevated privileges. 8. Ensure the documented security claims are supported by auditable implementation and automated tests.
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Rp1

Medium
Category
MCP Rug Pull
Confidence
93% confidence
Finding
The skill instructs users to run `npx openclaw` without pinning an exact package version, which allows whatever version is currently published or resolved at execution time to run. Because this is a security-themed skill that users may trust with elevated confidence, a compromised, typosquatted, or newly malicious package release could execute arbitrary code on the local system under the guise of auditing or hardening.

Static analysis

No suspicious patterns detected.