Security audit

EasyEDA API Skill

Security checks across malware telemetry and agentic risk

Overview

The skill is a legitimate EasyEDA automation/reference package, but it exposes a broad unauthenticated local bridge that can run code inside a live EasyEDA client and lacks strong user-confirmation guardrails for destructive or transactional actions.

Install only if you want an AI agent to actively control a running EasyEDA client. Keep the bridge stopped when not in use, only run it for trusted prompts and projects, review generated code before execution, and require explicit confirmation before deleting/modifying design data, switching projects, exporting files, or opening manufacturing/order flows.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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 (31)

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: The documented class exposes direct commercial ordering capabilities alongside export/reference functionality, including PCB, component, SMT, and 3D shell ordering. In an agent skill context, this broadens the action surface from design assistance to real-world procurement, which can lead to unintended purchases or workflow-triggering side effects if the agent invokes these methods without explicit user consent.

Context-Inappropriate Capability

Medium

Confidence: 96% confidence
Finding: place3DShellOrder can trigger ordering behavior and open a new order page, moving beyond passive design/debug capabilities into transactional actions. Even if the current implementation is partially limited, exposing this method to an agent creates a path for unauthorized or confusing procurement actions with external side effects.

Context-Inappropriate Capability

Medium

Confidence: 96% confidence
Finding: placeComponentsOrder enables component purchasing, which is materially different from reference or design automation and can produce business or financial consequences. In an agent environment, this creates risk of unapproved procurement, supplier interaction, or misleading automation if invoked based on ambiguous prompts.

Context-Inappropriate Capability

Medium

Confidence: 97% confidence
Finding: placePcbOrder can initiate PCB ordering, which has direct commercial and operational consequences beyond normal EDA assistance. Because PCB fabrication may incur cost, transmit manufacturing data, and launch external order flows, an agent with access to this method could cause unauthorized production actions from a simple design-related conversation.

Context-Inappropriate Capability

Medium

Confidence: 96% confidence
Finding: placeSmtComponentsOrder exposes SMT procurement/assembly ordering behavior, expanding the skill from documentation/design support into transactional manufacturing actions. This is dangerous because an agent may trigger costly assembly workflows or external page launches without the user's informed approval.

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: The documentation explicitly states that a caller can retrieve shortcut-key data for other extensions by constructing IDs with another extension's UUID prefix. Cross-extension introspection broadens the trust boundary and can enable enumeration of another extension's behavior or preparation for later interference, which exceeds a narrowly scoped design/debugging API.

Context-Inappropriate Capability

High

Confidence: 99% confidence
Finding: The API allows unregistering shortcut keys belonging to other extensions by supplying their prefixed IDs. This creates a direct integrity risk because one extension can disable another extension's functionality or user controls, enabling denial of service or facilitating UI manipulation against the user or competing extensions.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The README explicitly promotes direct source-file modification and a bridge that executes JavaScript in a running EasyEDA client, but it does not warn about destructive changes, project corruption, or abuse if an AI agent or local process submits unsafe code. In a skill package intended for AI coding tools, omission of safety boundaries materially increases the chance of unsafe autonomous actions and unintended code execution against live design data.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The README explicitly instructs users and agents to POST arbitrary JavaScript to a local bridge endpoint for execution inside the running EasyEDA client, but it does not warn that this can modify projects, corrupt design data, exfiltrate local design contents, or trigger unsafe actions in the EDA environment. In a skill intended for AI agents, omission of those safety boundaries is risky because an agent may treat the endpoint as routine tooling and execute generated code without adequate confirmation or sandboxing.

Vague Triggers

High

Confidence: 97% confidence
Finding: The trigger list includes very broad terms such as "EasyEDA", "PCB", "schematic", "footprint", and "EDA", which overlap with common technical vocabulary and can cause the skill to activate in unrelated contexts. Because this skill includes instructions to start local services, scan ports, and execute code in a connected EasyEDA client, accidental invocation expands the attack surface and may lead to unintended code execution workflows.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The documentation explicitly allows embedded images and files that may be downloaded or displayed inline, but it provides no warning or constraints around untrusted content. In an EDA tool and extension ecosystem, embedded objects can come from shared projects or external inputs, so rendering or exposing attachments without guidance on sanitization, content-type restrictions, and safe handling increases the risk of active-content abuse, phishing, or unsafe file delivery.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The example explicitly permits `data:text/html,<html></html>` as inline content, which normalizes embedding HTML via data URLs without any safety warning. If a client, plugin, or bridge renders such content in a web-capable context, this can enable script execution, credential theft, UI spoofing, or local privilege impacts depending on the host environment.

Missing User Warnings

Medium

Confidence: 81% confidence
Finding: The documentation exposes an API that captures the current rendered canvas as an image Blob but does not describe any privacy, consent, retention, or data-handling expectations. In the context of an agent skill with a WebSocket bridge into a live EasyEDA client, this omission increases the risk that agents or extensions could exfiltrate sensitive PCB/schematic content, proprietary designs, or embedded annotations without clear safeguards.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The documentation states that deleting a PCB may also delete the associated schematic and reusable module symbol, which creates a real risk of cascading data loss. In the context of an agent-executable EDA API, this is more dangerous because an agent or automation could invoke deletion without clearly surfacing the side effects to the user, leading to unintended project destruction.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The documentation explicitly states that calling openProject will directly discard unsaved changes in another already-open project, but it does not recommend a confirmation prompt, save check, or other safeguard. In the context of an agent skill that can drive a live EasyEDA client, this creates a real integrity and availability risk because an automated or mistaken invocation could cause irreversible user data loss.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The documentation exposes destructive deletion APIs for schematics and pages, including noted cascading deletion of related PCB data and reusable module symbols, without prominently warning callers about irreversible data loss or recommending confirmation/backup flows. In an agent skill context, this increases the chance that an AI agent could invoke deletion operations on behalf of a user without adequately surfacing the consequences.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The documentation states that disabling interactive checks can suppress in-app prompts and still open a new ordering page if conditions are met, but it does not present this as a prominent user safety warning. In an agent setting, inadequate warning increases the likelihood that developers expose or use the API in automation without realizing it can perform externally visible transactional actions silently.

Missing User Warnings

Medium

Confidence: 81% confidence
Finding: The documentation exposes a destructive delete capability without emphasizing confirmation, scoping, or safe-use guidance. In an agent skill context that can drive a live EasyEDA client, weak guardrails increase the risk of accidental or overly broad component deletion, causing integrity loss in PCB designs.

Missing User Warnings

Medium

Confidence: 85% confidence
Finding: The documented `delete` API performs destructive operations on PCB primitives, but the documentation provides no guidance to require user confirmation, scoping, preview, or undo-safe workflows. In an AI-agent context, that omission increases the chance that an agent could delete design elements automatically or from ambiguous prompts, causing integrity loss in engineering artifacts.

Missing User Warnings

Medium

Confidence: 82% confidence
Finding: The documented delete() API performs destructive design changes and the documentation lacks a clear warning about irreversible data loss, confirmation expectations, or backup/undo guidance. In the context of an AI-agent skill that can drive a live EasyEDA client, this increases the risk that an agent or user triggers destructive actions on PCB data without adequate safeguards.

Missing User Warnings

Medium

Confidence: 80% confidence
Finding: The modify() API can change PCB geometry, layer placement, connectivity metadata, and lock state, yet the documentation does not warn about integrity impact, validation needs, or potential manufacturing/design consequences. Because this skill is intended for agent-driven interaction with a running EDA environment, under-documented mutation APIs are more dangerous than in static reference material alone.

Missing User Warnings

Medium

Confidence: 85% confidence
Finding: The documented behavior allows non-interactive execution to silently pass checks and open an ordering page, and `ignoreWarning=true` can suppress all warnings. In an agent skill context, this can enable unintended procurement actions or workflow manipulation without meaningful user awareness, increasing the risk of accidental purchases or submission of incomplete manufacturing data.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: This SMT ordering method has the same silent non-interactive behavior and warning-bypass option, creating a real risk in an AI-agent environment where actions may be triggered automatically from natural-language requests. An agent could place or advance component ordering flows without clear user intent, especially if warnings are ignored programmatically.

Missing User Warnings

Medium

Confidence: 79% confidence
Finding: The delete() API performs destructive operations on schematic primitives, but the documentation does not warn that calling it can permanently alter or remove user design data. In a skill ecosystem for AI agents that may autonomously invoke APIs, insufficient warning increases the risk of unintended destructive actions, especially when users may not realize the operation is irreversible or should require confirmation.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The documentation exposes high-impact capabilities to import projects and modify document source, but it does not clearly warn that these operations can overwrite, alter, or introduce data into active user projects. In an agent skill context, this omission is dangerous because an LLM-driven agent may treat these actions as routine and perform destructive changes or untrusted imports without explicit user confirmation.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.