Wormxgpt MCP Server

This server (without client built-ins) exposes a complete data-exfiltration chain: get_github_repo → get_env_vars ⇒ http_request. Untrusted input is ingested, private data is read, and it can be sent to an external sink — and at least one leg is a direct schema wire (⇒), where a producer's output drops straight into a free-text parameter of the next tool, so the chain needs little agent cooperation. Static analysis proves the primitive exists, not that a specific run will occur.

Fix: Remove one leg of the trifecta: isolate untrusted-input tools from secret-reading tools and from egress tools, or require human approval between them.

Location: flow get_github_repo → get_env_vars → http_request

Tool "run_shell_command" appears to run shell commands or evaluate code (keyword "shell" in tool name, parameter "command"). Arbitrary execution driven by model input is one of the most dangerous MCP capabilities; combined with any untrusted input it becomes RCE.

Fix: Sandbox execution, allowlist commands/arguments, and never pass model output to a shell unescaped.

Location: tool run_shell_command

Tool "run_puppeteer_script" appears to run shell commands or evaluate code (parameter "script"). Arbitrary execution driven by model input is one of the most dangerous MCP capabilities; combined with any untrusted input it becomes RCE.

Fix: Sandbox execution, allowlist commands/arguments, and never pass model output to a shell unescaped.

Location: tool run_puppeteer_script

In the server's implementation (`cli/customTools.ts:41`): Evaluating strings as code is the most direct RCE primitive; if any tool input reaches it, the server executes attacker-chosen code. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: nction const fn = new Function('args', ` return (async () => { ${t.code} })(); `);

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server cli/customTools.ts

In the server's implementation (`services/tools.ts:7235`): Evaluating strings as code is the most direct RCE primitive; if any tool input reaches it, the server executes attacker-chosen code. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const result = Function(`"use strict"; return (${sanitized})`)(); return `${args.expression} = ${resul

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server services/tools.ts

In the server's implementation (`bin/wormxgpt.js:11`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: import { spawn } from 'child_process'; const __filename = fileURLToPath(import.meta.url); const __dirname = dirname(__

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server bin/wormxgpt.js

In the server's implementation (`cli/index.ts:855`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: spawn } = await import('child_process'); const serverPath = path.join(path.dirname(new URL(import.meta.url).path

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server cli/index.ts

In the server's implementation (`mcp-server.js:70`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: require("crypto"); var child_process_1 = require("child_process"); var app = (0, express_1.default)(); app.use(express_

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server mcp-server.js

In the server's implementation (`mcp-server.ts:12`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: port { execSync } from "child_process"; import puppeteerExtra from "puppeteer-extra"; import StealthPlugin from "puppe

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server mcp-server.ts

In the server's implementation (`services/tools.ts:7208`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: cSync } = await import('child_process'); const out = execSync(args.command, { cwd: args.cwd, timeout: 15000, en

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server services/tools.ts

In the server's implementation (`utils/workspaceContext.ts:3`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: port { execSync } from 'child_process'; export interface WorkspaceSummary { fileCount: number; gitBranch: strin

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server utils/workspaceContext.ts

In the server's implementation (`cli/doctor.ts:19`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: ; const res = await fetch('https://registry.npmjs.org/', { method: 'HEAD' }); const ms = Date.now() - start;

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server cli/doctor.ts

In the server's implementation (`mcp-server.js:579`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: 0: return [4 /*yield*/, fetch("https://dns.google/resolve?name=".concat(encodeURIComponent(a.domain), "&type=").concat(t

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server mcp-server.js

In the server's implementation (`mcp-server.ts:624`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const r = await fetch(`https://dns.google/resolve?name=${encodeURIComponent(a.domain)}&type=${t}`); c

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server mcp-server.ts

In the server's implementation (`services/cohere.ts:32`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const response = await fetch('https://api.cohere.com/v2/models', { headers: { 'Authorization': `Bearer ${key}`

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server services/cohere.ts

In the server's implementation (`services/integrations.ts:415`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: se<string> { await fetch(`https://api.github.com/user/starred/${owner}/${repo}`, { method: 'PUT', hea

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server services/integrations.ts

In the server's implementation (`services/moonshot.ts:14`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const response = await fetch("https://api.moonshot.ai/v1/chat/completions", { method: 'POST', headers

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server services/moonshot.ts

In the server's implementation (`services/openai.ts:192`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const response = await fetch('https://api.openai.com/v1/models', { headers: { 'Authorization': `Bea

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server services/openai.ts

In the server's implementation (`services/tools.ts:248`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: { const r = await fetch(`https://translate.googleapis.com/translate_a/single?client=gtx&sl=${sl}&tl=${tl}&dt=t&q=$

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server services/tools.ts

In the server's implementation (`services/voiceMode.ts:157`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: const response = await fetch('https://gen.pollinations.ai/v1/audio/transcriptions', { method: 'POST',

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server services/voiceMode.ts

Tool "http_request" takes a URL/host parameter "url" with no allowlist/pattern. An outbound-request tool with an unbounded destination enables SSRF and cloud-metadata access (e.g. 169.254.169.254).

Fix: Allowlist destinations or constrain the parameter; block private/link-local addresses server-side.

Location: tool http_request · inputSchema.properties.url

Tool "write_file" can write, overwrite or delete files (keyword "write_file" in tool name). Verify it is scoped to a safe directory.

Fix: Constrain file operations to an explicit, non-sensitive root; reject path traversal.

Location: tool write_file

Tool "delete_file" can write, overwrite or delete files (keyword "delete_file" in tool name). Verify it is scoped to a safe directory.

Fix: Constrain file operations to an explicit, non-sensitive root; reject path traversal.

Location: tool delete_file

Tool "run_shell_command" takes a command-shaped parameter "command" with no enum/pattern constraint. Free-form, model- or attacker-controlled arguments reaching a shell is the command-injection precondition.

Fix: Constrain the parameter (enum/pattern), or build the command from a fixed template with escaped args.

Location: tool run_shell_command · inputSchema.properties.command

Tool "run_puppeteer_script" takes a command-shaped parameter "script" with no enum/pattern constraint. Free-form, model- or attacker-controlled arguments reaching a shell is the command-injection precondition.

Fix: Constrain the parameter (enum/pattern), or build the command from a fixed template with escaped args.

Location: tool run_puppeteer_script · inputSchema.properties.script

Tool "write_file" can mutate/egress but declares no destructiveHint. Clients that don't default to spec-safe behavior may not prompt before running it.

Fix: Declare accurate annotations, and gate destructive tools on user confirmation regardless.

Location: tool write_file

Tool "delete_file" can mutate/egress but declares no destructiveHint. Clients that don't default to spec-safe behavior may not prompt before running it.

Fix: Declare accurate annotations, and gate destructive tools on user confirmation regardless.

Location: tool delete_file

Tool "run_shell_command" can mutate/egress but declares no destructiveHint. Clients that don't default to spec-safe behavior may not prompt before running it.

Fix: Declare accurate annotations, and gate destructive tools on user confirmation regardless.

Location: tool run_shell_command

Tool "run_puppeteer_script" can mutate/egress but declares no destructiveHint. Clients that don't default to spec-safe behavior may not prompt before running it.

Fix: Declare accurate annotations, and gate destructive tools on user confirmation regardless.

Location: tool run_puppeteer_script

Tool "read_file" takes a path parameter "path" with no constraint. Without a canonicalize-and-contain check (not visible statically), this permits ../ traversal outside the intended root.

Fix: Resolve and verify the path stays within an allowed root; reject traversal sequences.

Location: tool read_file · inputSchema.properties.path

Tool "write_file" takes a path parameter "path" with no constraint. Without a canonicalize-and-contain check (not visible statically), this permits ../ traversal outside the intended root.

Fix: Resolve and verify the path stays within an allowed root; reject traversal sequences.

Location: tool write_file · inputSchema.properties.path

Tool "delete_file" takes a path parameter "path" with no constraint. Without a canonicalize-and-contain check (not visible statically), this permits ../ traversal outside the intended root.

Fix: Resolve and verify the path stays within an allowed root; reject traversal sequences.

Location: tool delete_file · inputSchema.properties.path