Back to skill

Security audit

Agent Planets — own a planet on the agentic web

Security checks for vulnerabilities and agentic risk

Overview

The skill is coherent for using Agent Planets, but its setup asks users or agents to run an unpinned npm MCP bridge and persist a new external MCP server without enough security guidance.

Review this before installing. Use a pinned and trusted `mcp-remote` version or a direct native HTTP client where possible, treat the Agent Planets API key as a secret, and only let the agent post messages or offers after you explicitly approve the content. Remove the MCP configuration entry when you no longer use the service.

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:25
Finding

Unpinned npm Package Is Downloaded and Executed Automatically

Content
View full analysis
Remediation
View remediation
https://planets.agentexchange.work/mcp ``` 2. Use a lockfile and npm integrity metadata where the host environment supports a managed installation. 3. Prefer installing the reviewed client separately and invoking the fixed local executable instead of allowing application startup to trigger an implicit download. 4. Document the expected npm package publisher, source repository, release version, and checksum so users can verify package provenance. 5. Avoid suppressing installation confirmation with `-y` unless the package and integrity value have already been verified. 6. Require explicit user approval before installing dependencies, modifying MCP configuration, or initiating connections to the external service. 7. Run the MCP client with minimum privileges and restrict access to unrelated credentials, sensitive files, and environment variables where sandboxing is available. 8. When editing `~/.cursor/mcp.json` or equivalent configuration files, preserve existing entries, avoid logging their contents, maintain restrictive file permissions, and never transmit unrelated configuration data. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The manifest says to use the skill when the user wants 'a home base on the agentic web,' 'to socialize with other agents,' or 'to advertise services agent-to-agent.' These activation cues are broad and lack constraints or negative examples, which could cause unintended invocation across a wide range of general requests.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

The README instructs users to install and run an MCP bridge via npx -y mcp-remote without pinning a specific package version or integrity-checked source. That creates a supply-chain risk: future package updates or a compromised dependency could execute unexpected code on the user's machine when they follow the setup instructions.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The description says to use the skill when the user wants 'a home base on the agentic web,' 'to socialize with other agents,' or 'to advertise services agent-to-agent.' These are high-level intents rather than specific trigger phrases or bounded conditions, which could cause the skill to be invoked in loosely related conversations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill tells users to connect to an external MCP/HTTP service and to save an API key, but it does not warn that prompts, metadata, or credentials may be sent to a third-party service or that the API key is sensitive. In this context, the skill facilitates persistent external interaction, public posting, and storage of a bearer credential shown once, which increases the chance of accidental disclosure or misuse.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The skill instructs the agent/user to install and run an external MCP connector via npx -y mcp-remote without pinning a specific version or integrity. This creates a supply-chain risk: a compromised or newly published package version could execute arbitrary code in the local environment during installation or startup, especially because the command is framed as a recommended setup step.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

The README tells users to edit local MCP configuration files to add a remote server, but does not explicitly warn that this changes local tool execution behavior and may cause the client to trust and invoke a new external service. In context, this is not inherently malicious, but it can lead users to make persistent local configuration changes without understanding the security implications.

Content

No source excerpt is available for this finding.