Indusk MCP Server

Tool "quality_check" appears to run shell commands or evaluate code (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 quality_check

In the server's implementation (`admin/.next/build/chunks/node_modules__pnpm_0e-dfa2._.js:2548`): 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: ); let templateFn = new Function(`return ${template}`)(); return templateFn(comparator); } /** * Sort the given

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 admin/.next/build/chunks/node_modules__pnpm_0e-dfa2._.js

In the server's implementation (`admin/.next/server/chunks/ssr/[root-of-the-server]__0p.51_4._.js:1`): 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: +str.trim()+";\n}());"),eval(str)||{}}catch(err){if(!1!==wrap&&/(unexpected|identifier)/i.test(err.message))return parse

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 admin/.next/server/chunks/ssr/[root-of-the-server]__0p.51_4._.js

In the server's implementation (`hooks/eval-trigger.js:186`): 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: ined ${drained} pending eval(s); ${failed.length} FAILED and remain queued for retry; ${alreadyDrained} already drained.

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 hooks/eval-trigger.js

In the server's implementation (`admin/.next/build/chunks/[root-of-the-server]__0-n8fv.._.js:2`): 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: s = [ "[turbopack-node]/child_process/globals.ts [postcss] (ecmascript)", ((__turbopack_context__, module, exports) => {

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 admin/.next/build/chunks/[root-of-the-server]__0-n8fv.._.js

In the server's implementation (`admin/.next/build/postcss.js:4`): 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: ) R.m("[turbopack-node]/child_process/globals.ts [postcss] (ecmascript)") R.m("[turbopack-node]/child_process/evaluate.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 admin/.next/build/postcss.js

In the server's implementation (`admin/.next/server/chunks/ssr/[root-of-the-server]__0f9b~qs._.js:1`): 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: )=>{b.exports=a.x("node:child_process",()=>require("node:child_process"))},54779,a=>{"use strict";var b=a.i(44673),c=a.i

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 admin/.next/server/chunks/ssr/[root-of-the-server]__0f9b~qs._.js

In the server's implementation (`dist/bin/commands/agent.js:1`): 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: spawnSync } from "node:child_process"; import { existsSync, readFileSync, renameSync, writeFileSync } from "node:fs"; i

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 dist/bin/commands/agent.js

In the server's implementation (`dist/bin/commands/doppler.js:1`): 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: spawnSync } from "node:child_process"; import { existsSync, readdirSync, readFileSync, statSync, writeFileSync } from "

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 dist/bin/commands/doppler.js

In the server's implementation (`dist/bin/commands/eval.js:152`): 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: } = await import("node:child_process"); const { rmSync, writeFileSync } = await import("node:fs"); const { base

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 dist/bin/commands/eval.js

In the server's implementation (`dist/bin/commands/extensions.js:1`): 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: { execSync } from "node:child_process"; import { cpSync, existsSync, mkdirSync, readdirSync, readFileSync, rmSync, write

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 dist/bin/commands/extensions.js

In the server's implementation (`dist/bin/commands/infra.js:1`): 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: { execSync } from "node:child_process"; import { existsSync, mkdirSync, readFileSync, writeFileSync } from "node:fs"; im

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 dist/bin/commands/infra.js

In the server's implementation (`dist/bin/commands/init-docs.js:1`): 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: { execSync } from "node:child_process"; import { existsSync, mkdirSync, readFileSync, writeFileSync } from "node:fs"; im

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 dist/bin/commands/init-docs.js

In the server's implementation (`dist/bin/commands/init.js:1`): 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: spawnSync } from "node:child_process"; import { cpSync, existsSync, mkdirSync, readFileSync, writeFileSync } from "node

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 dist/bin/commands/init.js

In the server's implementation (`dist/server/index.js:35`): 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: s(currentVersion) { fetch("https://registry.npmjs.org/@infinitedusky/indusk-mcp/latest") .then((res) => res.

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 dist/server/index.js

In the server's implementation (`dist/bin/commands/eval.js:167`): A shell/process command assembled from concatenated or interpolated values is command injection when any part is attacker-influenced — the OWASP canonical RCE flow. Verify what reaches the interpolated value. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: HEAD try { execSync(`git worktree add "${worktreePath}" HEAD --detach`, { cwd: projectRoot,

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 dist/bin/commands/eval.js

In the server's implementation (`dist/bin/commands/extensions.js:188`): A shell/process command assembled from concatenated or interpolated values is command injection when any part is attacker-influenced — the OWASP canonical RCE flow. Verify what reaches the interpolated value. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: e: true }); execSync(`npm pack ${pkg} --pack-destination "${tmpDir}"`, { encoding: "utf-8",

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 dist/bin/commands/extensions.js

In the server's implementation (`dist/tools/graph-tools.js:31`): A shell/process command assembled from concatenated or interpolated values is command injection when any part is attacker-influenced — the OWASP canonical RCE flow. Verify what reaches the interpolated value. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: t to Redis port execSync(`node -e "const s=require('net').connect(6379,'${host}');s.setTimeout(2000);s.on('conne

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 dist/tools/graph-tools.js

Tool "quality_check" 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 quality_check · inputSchema.properties.command

In the server's implementation (`admin/.next/build/chunks/[turbopack]_runtime.js:588`): Loading a module chosen at runtime (from a variable) can pull in and run attacker-influenced code paths. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: y { raw = await import(id); } catch (err) { // TODO(alexkirsz) This can happen when a client-side mo

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 admin/.next/build/chunks/[turbopack]_runtime.js

In the server's implementation (`admin/.next/server/chunks/[turbopack]_runtime.js:601`): Loading a module chosen at runtime (from a variable) can pull in and run attacker-influenced code paths. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: y { raw = await import(id); } catch (err) { // TODO(alexkirsz) This can happen when a client-side mo

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 admin/.next/server/chunks/[turbopack]_runtime.js

In the server's implementation (`admin/.next/server/chunks/ssr/[turbopack]_runtime.js:601`): Loading a module chosen at runtime (from a variable) can pull in and run attacker-influenced code paths. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: y { raw = await import(id); } catch (err) { // TODO(alexkirsz) This can happen when a client-side mo

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 admin/.next/server/chunks/ssr/[turbopack]_runtime.js

In the server's implementation (`hooks/eval-trigger.js:313`): Loading a module chosen at runtime (from a variable) can pull in and run attacker-influenced code paths. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: olvedFindings } = await import(findingsPath); const unresolved = getUnresolvedFindings(statePath); if (unresolved.le

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 hooks/eval-trigger.js

Tool "quality_check" 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 quality_check