Security audit

Deep Research

Security checks for vulnerabilities and agentic risk

Overview

The skill appears intended for AnyGen deep research, but it installs and delegates to unpinned external components that can change after review.

Review this skill before installing. It is suitable only if you trust AnyGen, the npm package source for @anygen/cli, and the remote anygen-workflow-generate workflow. Avoid submitting secrets or confidential research topics unless you are comfortable sending them to AnyGen, and prefer a pinned or audited workflow version without unattended installation.

A.I.G

Vulnerability Patterns

Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
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
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally

Findings (2)

T08 · Insecure Dependencies

Warning

Location: SKILL.md:11
Finding: Unpinned Executable npm Dependency
Content: View full analysis
Remediation: View remediation

T03 · Remote Payload Retrieval and Execution

Error

Location: SKILL.md:41
Finding: Unverified Runtime Installation of an External Skill
Content: View full analysis
-y ``` ``` ### Technical Analysis The reviewed package delegates its primary behavior to `anygen-workflow-generate`, but that workflow is not included in the audited project. If it is unavailable, the skill directs the agent to install it dynamically through the AnyGen CLI. The installation command does not identify an immutable workflow version, source revision, or integrity digest. The `-y` option also indicates unattended confirmation, reducing the opportunity for a user to inspect and approve the retrieved content. As a result, the effective instructions executed by the agent can differ from the content reviewed in `SKILL.md`. If the remote artifact or its distribution channel is compromised, attacker-controlled instructions or executable components could be introduced after the original skill audit. No evidence establishes that the current external workflow or AnyGen service is malicious. The finding concerns the unaudited, mutable remote trust boundary. ### Attack Path 1. An attacker compromises the remote skill artifact, its publisher, or the service used to distribute it. 2. The attacker replaces or updates `anygen-workflow-generate` with malicious instructions or executable content. 3. An agent loads this skill in an environment where the delegated workflow is absent. 4. Following `SKILL.md`, the agent runs the unattended `anygen skill install` command. 5. The mutable external workflow is downloaded and installed without a pinned version or integrity verification. 6. The agent subsequently follows the attacker-cont ...[truncated 669 chars]
Remediation: View remediation

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The skill’s trigger criteria are extremely broad and include many common requests for analysis or research, making accidental invocation likely. This can route ordinary user queries into a tool that performs external operations and may send user content to a third-party service, increasing privacy, cost, and unintended tool-use risk.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill explicitly states that research is generated server-side at an external domain, but it does not warn that user prompts, research topics, and potentially sensitive content will be transmitted to that third-party service. In a research skill, users may provide confidential business, regulatory, investment, or proprietary information, so lack of disclosure creates a meaningful data exposure risk.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.