Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed xAI/Grok API wrapper for web/X search plus optional chat, vision, and model listing, with no evidence of hidden persistence or destructive behavior.
Install only if you are comfortable sending prompts, search queries, filters, and any explicitly provided images to xAI using your XAI_API_KEY. Avoid using it with secrets, private documents, or sensitive images unless that third-party API use is acceptable, and consider narrowing triggers if accidental invocation would be a problem.
The skill invokes bundled Node scripts that require environment access and make outbound API calls, yet the manifest does not declare any explicit tool scope such as permissions or allowed-tools. This creates a transparency and policy-enforcement gap: an agent or reviewer may treat the skill as lower risk than it is, increasing the chance of unintended secret access or external data transmission.
The manifest presents the skill as search-oriented, but the documentation exposes materially broader capabilities: freeform chat, vision chat, and model listing. This scope mismatch can mislead operators and upstream agents into routing arbitrary prompts or images to an external model when they only intended web/X search, increasing data-exposure and misuse risk.
The skill does not clearly warn users that their queries, prompts, and possibly image inputs are transmitted to the external xAI Responses API. Without an explicit disclosure, users or calling agents may send sensitive data under the false assumption that processing is local, leading to avoidable third-party data exposure.
Shadow Command Trigger: 'search x' conflicts with built-in command 'search'
Shadow Command Trigger: 'search twitter' conflicts with built-in command 'search'
The manifest trigger list includes generic phrases such as "grok", "xai", "find tweets", and "twitter search" without any exclusion conditions or scope constraints. These are broad enough to match ordinary user language and could cause unintended invocation of this skill.
Shadow Command Trigger: 'find tweets' conflicts with built-in command 'find'
The implementation materially exceeds the declared skill scope: instead of a constrained search tool using server-side web_search/x_search, it provides general-purpose Grok chat. Scope drift is dangerous because users and orchestrators may trust the manifest and route broader prompts or sensitive content into a capability they did not intend to expose, undermining least privilege and policy enforcement.
The script adds local image ingestion and uploads image contents to the remote API, even though the skill is described as a web/X search utility. This expands the data-exfiltration surface beyond user expectations and can cause accidental transmission of sensitive local files if an operator assumes the skill only sends search queries.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
store: false,
};
const res = await fetch("https://api.x.ai/v1/responses", {
method: "POST",
headers: {
"Content-Type": "application/json",
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
store: false,
};
const res = await fetch("https://api.x.ai/v1/responses", {
method: "POST",
headers: {
"Content-Type": "application/json",
This code transmits the user's query, selected search mode, and filtering parameters to an external network service via an HTTP POST request. Although the file header describes the tool as an xAI search wrapper, there is no runtime warning, confirmation, or explicit user-facing notice that the provided query data will be sent off-box to a third-party API.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
let data;
try {
data = await fetchJson("https://api.x.ai/v1/language-models");
} catch {
data = await fetchJson("https://api.x.ai/v1/models");
}
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
let data;
try {
data = await fetchJson("https://api.x.ai/v1/language-models");
} catch {
data = await fetchJson("https://api.x.ai/v1/models");
}
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
let data;
try {
data = await fetchJson("https://api.x.ai/v1/language-models");
} catch {
data = await fetchJson("https://api.x.ai/v1/models");
}
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
let data;
try {
data = await fetchJson("https://api.x.ai/v1/language-models");
} catch {
data = await fetchJson("https://api.x.ai/v1/models");
}
Detected: suspicious.dangerous_exec, suspicious.env_credential_access, suspicious.exposed_secret_literal