Back to skill

Security audit

A2A Marketplace

Security checks for vulnerabilities and agentic risk

Overview

This skill is a marketplace connector that openly supports paid third-party tool execution, but it lacks enough safeguards and install pinning for that high-impact behavior.

Review this carefully before installing. Use it only if you trust the A2A/AgentForge plugin publisher and are comfortable with marketplace tools receiving submitted inputs and incurring charges. Check balance and pricing before execution, avoid sensitive data unless provider handling is clear, and prefer a pinned, reviewed plugin version with explicit spending controls.

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:10
Finding
Unpinned Third-Party Node Package Creates a Supply-Chain Risk## Vulnerability Details **File Location**: `SKILL.md`, lines 10–15 **Vulnerability Type**: `T08: Insecure Dependencies` **Risk Level**: Medium ### Vulnerable Code ```yaml { "id": "plugin", "kind": "node", "package": "@a2a/openclaw-plugin", "label": "Install A2A Corp plugin", } ``` ### Technical Analysis The skill directs the platform to install the third-party Node package `@a2a/openclaw-plugin` without specifying an exact version or package integrity hash. The project also contains no lockfile or vendored plugin source that would allow the installed artifact to be compared with the version reviewed during the audit. As a result, package resolution may retrieve a release that differs from the one originally intended or reviewed. The effective executable code can therefore change after this skill file has been approved. This exposes installations to package-publisher compromise, registry-account compromise, malicious future releases, and unintended behavior introduced by dependency updates. The repository contains only `SKILL.md`; the plugin implementation was not available for inspection. No malicious payload was identified in the audited file itself. ### Attack Path 1. An attacker compromises the package publisher account, publication token, registry channel, or another relevant part of the package supply chain. 2. The attacker publishes a malicious or compromised release of `@a2a/openclaw-plugin`. 3. A user or agent installs the skill after that release becomes the version selected by the package resolver. 4. Because the dependency has no exact version or verified integrity value, the installation retrieves the changed package. 5. The package's installation or runtime code executes within the permissions granted to the host agent or Node process. 6. The malicious code may inspect agent inputs, manipulate marketplace requests, interfere with billed operations, access resources available to the process, or perform other actions permitted by the ...[truncated 752 chars]
Remediation
## Remediation Suggestions 1. Pin `@a2a/openclaw-plugin` to a reviewed, immutable, exact version rather than relying on mutable package resolution. 2. Record and enforce the expected package integrity hash through an appropriate lockfile or installation mechanism. 3. Commit the lockfile to the project and require deterministic, locked dependency installation. 4. Verify package publisher ownership, registry provenance, and release signatures or attestations where supported. 5. Review the plugin source and its transitive dependencies before approval, because they were not included in this repository. 6. Use automated dependency monitoring and require security review before updating the pinned version. 7. Run the plugin in a least-privilege sandbox with restricted filesystem, environment-variable, credential, and network access. 8. Require explicit authorization and spending limits for billed marketplace executions so dependency compromise cannot silently create unrestricted charges.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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)

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The skill explicitly encourages discovering and executing third-party marketplace tools with 'billing applied automatically' but provides no warning that user inputs may be transmitted to external tools or that execution can incur real spending. This creates a meaningful risk of unintended charges and inadvertent disclosure of sensitive data, especially because the documentation normalizes execution as a routine workflow step.

Static analysis

No suspicious patterns detected.