Is relay-companion safe?
- Defense-evasion command idiom
- Node.js shell/command execution
- Node.js filesystem read
What to do: Nothing here argues against installing it. Grant the capabilities it lists only if you expect the tool to need them.
relay-companion is an npm package analyzed by SkillTotal's deterministic static scanner. The scan found no malicious indicators, though 1 risky construct is reported for review. It can: delegated authentication, filesystem read, filesystem write, mcp tools detected, network egress and shell execution — capabilities are what the code can do, not a verdict on intent. Risk score 20/100 (low).
relay-companion 0.1.440
Get notified when this component changes
Change alerts aren't live yet. Leave your email and we'll let you know when they launch. We use it only for this and delete it on request.
Automated static-analysis result. It can contain false positives and false negatives, and is not a claim about the intent of relay-companion's authors. Report a false positive.
Behavioral traits
How this component maps to the CSA agentic threat model. Descriptive — it never affects the risk score.
Findings (7)
A command uses a known defense-evasion idiom: PowerShell execution-policy bypass / encoded command / hidden window, macOS code-signing bypass, or launching a payload from a world-writable temp directory. These are hallmarks of droppers and rarely appear in legitimate code.
const signed = runCommand("/usr/bin/codesign", ["--force", "--deep", "--sign", "-", tmpPath]);Fix: Verify why the component bypasses execution policy / code signing or runs from a temp directory; these patterns are characteristic of malware staging.
The component can run operating-system commands or spawn processes.
import { spawnSync, spawn } from "node:child_process";const result = spawnSync(command, args, { stdio: "ignore" });const beforePidResult = spawnSync("systemctl", ["--user", "show", "--property=MainPID", "--value", "work.relay.companion.pill.service"], {const mainPid = spawnSync("systemctl", ["--user", "show", "--property=MainPID", "--value", "work.relay.companion.pill.service"], {const child = spawn(
const du = spawnSync("du", ["-sk", layout.releasesDir], { encoding: "utf8", timeout: 30_000 });const probe = spawnSync(
const mime = spawnSync("xdg-mime", ["query", "default", "x-scheme-handler/relay"], {spawnSync("launchctl", ["kickstart", "-k", `gui/${uid}/work.relay.companion.pill`], { stdio: "ignore" });const result = spawnSync(process.execPath, [bootstrap, "setup", ...rest], {const { spawnSync } = require("node:child_process");const { spawnSync } = require("node:child_process");const names = spawnSync(nameInvocation.command, nameInvocation.args, { cwd: nameInvocation.cwd, encoding: "utf8", windowsHide: true });const details = spawnSync(detailInvocation.command, detailInvocation.args, { cwd: detailInvocation.cwd, encoding: "utf8", windowsHide: true });const { spawnSync } = require("node:child_process");const result = spawnSync(String(current.node), [bin, ...process.argv.slice(2)], { stdio: "inherit", env: process.env, windowsHide: true });const unpack = spawnSync(unpackInvocation.command, unpackInvocation.args, {const { spawnSync } = require("node:child_process");return spawnSync(command, args, { encoding: "utf8", timeout: 30_000, windowsHide: true, ...options });const { spawn } = require("node:child_process");const cp = require("node:child_process");const child = spawn(
const child = spawn(
const child = spawn(
const child = spawn(cmd, args, { detached: true, stdio: "ignore" });Why it matters: Powerful and often legitimate — confirm the commands aren't built from untrusted input.
Fix: Confirm the command and its arguments are fully controlled and not derived from untrusted input; prefer execFile with an argument array.
The component reads files from disk.
return JSON.parse(fsSync.readFileSync(p, "utf8")) || {};const tail = fsSync.readFileSync(updateLogPath, "utf8");
try { unitNamesActiveRoot = fsSync.readFileSync(expectedPath, "utf8").includes(activeUnitRoot); } catch {}try { stale = !fsSync.readFileSync(file, "utf8").includes(activeDesktopRoot); }const input = fs.createReadStream(file);
try { document = JSON.parse(fs.readFileSync(mapPath, "utf8")); } catch {const manifest = JSON.parse(fs.readFileSync(path.join(packageRoot, "package.json"), "utf8"));
try { current = JSON.parse(fs.readFileSync(pointerPath, "utf8")); }const existingLauncher = fs.existsSync(launcherPath) ? fs.readFileSync(launcherPath, "utf8") : null;
const existingShim = fs.existsSync(shimPath) ? fs.readFileSync(shimPath, "utf8") : null;
overlayPrefs = JSON.parse(fs.readFileSync(OVERLAY_PREFS_PATH, "utf8")) || {};raw = fs.readFileSync(configPath, "utf8");
const ledger = JSON.parse(fs.readFileSync(ledgerPath, "utf8")) || {};raw = JSON.parse(fs.readFileSync(STATE_PATH, "utf8")) || {};const store = JSON.parse(fs.readFileSync(STATE_PATH, "utf8")) || {};return Number(JSON.parse(fs.readFileSync(metaPath, "utf8")).lastFocusedAt) || 0;
settings = JSON.parse(fs.readFileSync(install.claudeSettingsPath(), "utf8")) || {};targetCliPid = Number(JSON.parse(fs.readFileSync(rendezvousPath, "utf8")).cliPid) || 0;
const published = JSON.parse(fs.readFileSync(directory.sessionDirectoryStatePath(), "utf8"));
const raw = fs.readFileSync(SCHEDULES_PATH, "utf8");
const b = fs.readFileSync(file);
const serialized = fs.readFileSync(statePath, "utf8");
const value = JSON.parse(fs.readFileSync(file, "utf8"));
try { meta = JSON.parse(fs.readFileSync(metaPath, "utf8")); } catch {}const body = await fs.readFile(filePath);
Why it matters: Usually legitimate, but worth confirming it can't be steered into reading sensitive files.
Fix: Confirm which files are read and that paths cannot be influenced by untrusted input to reach sensitive locations.
The component writes or deletes files on disk.
fsSync.writeFileSync(prefsPath, `${JSON.stringify(prefs, null, 2)}\n`);const output = fs.createWriteStream(file, { flags: "r+", start: offset, mode: 0o600 });fs.rmSync(file, { force: true });fs.writeFileSync(temporary, `${JSON.stringify(value, null, 2)}\n`, { mode: 0o600 });fs.writeFileSync(temporary, value, { mode, flag: "wx" });try { fs.rmSync(temporary, { force: true }); } catch {}removePointer = (file) => fs.rmSync(file, { force: true }),try { fs.rmSync(failedRoot, { recursive: true, force: true }); } catch {}fs.rmSync(staging, { recursive: true, force: true });fs.writeFileSync(ledgerPath, `${JSON.stringify(ledger, null, 2)}\n`);fs.rmSync(claimed, { force: true });fs.rmSync(eventFile, { force: true });fs.writeFileSync(target, Buffer.from(encoded, "base64"), { mode: 0o600 });try { fs.rmSync(tempDir, { recursive: true, force: true }); } catch {}fs.writeFileSync(path.join(noteDir, `${safeNoteStem(id)}.md`), note);fs.writeFileSync(storedPath, bytes, { mode: 0o600 });fs.writeFileSync(imagePath, bytes, { mode: 0o600 });fs.writeFileSync(path.join(noteDir, `${safeNoteStem(id)}.md`), composerNote);fs.writeFileSync(SCHEDULES_PATH, JSON.stringify(all, null, 2));
fs.appendFileSync(PERF_LOG_PATH, `${line}\n`);fs.appendFileSync(
fs.writeFileSync(tmp, `${JSON.stringify(value, null, 2)}\n`, { mode: 0o600 });fs.rmdirSync(lock);
try { fs.rmdirSync(lock); } catch {}try { fs.unlinkSync(file); } catch {}Why it matters: Usually legitimate, but worth confirming the paths can't be controlled by untrusted input.
Fix: Confirm which files are written/deleted and that paths cannot be influenced by untrusted input.
The component makes outbound network requests.
const https = require("node:https");const request = https.get(parsed, { headers: { "user-agent": `${PACKAGE_NAME}/${packageJson.version}` } }, (response) => {res = await fetch(upload.url, { method: "POST", body: form });res = await fetch(upload.url, {return { dispatcher: null, fetch: (u, i) => fetch(u, i) };return transport.fetch(url, transport.dispatcher ? { ...init, dispatcher: transport.dispatcher } : init);import http from "node:http";
const req = http.get({ host: INSPECTOR_HOST, port, path: "/json/list", timeout: 300 }, (res) => {const response = await fetch(attachment.downloadUrl, { signal: AbortSignal.timeout(30_000) });import { createServer as createHttpServer } from "node:http";import { createServer as createHttpsServer } from "node:https";const res = await fetch(url, { signal: AbortSignal.timeout(30000) });Why it matters: Usually legitimate, but confirm the destinations are expected and no sensitive data leaves.
Fix: Confirm the destination hosts are expected and that no sensitive data is sent off-host.
An OAuth 2.0 / OpenID Connect delegated-authentication flow was detected (authorization-code / refresh-token / token-exchange grant, an OIDC authorize/discovery endpoint or id_token, or a delegation library). Tools authenticate with the end user's delegated, scoped credentials rather than a long-lived embedded service credential. (5 occurrence(s) shown as evidence).
authorization_endpoint: `${origin}/oauth/authorize`,if (req.method === "GET" && pathname === "/oauth/authorize") {html(res, 200, `<!doctype html><html lang="en"><meta charset="utf-8"><meta name="viewport" content="width=device-width"><title>Connect Claude to Relay</title><style>body{font:16px system-ui;max-width:32rem;margin:4rem auto;padding:1rem;line …if (req.method === "POST" && pathname === "/oauth/authorize") {if (provider === "claude" && parsed.hostname === "claude.com" && parsed.pathname === "/cai/oauth/authorize") {Fix: Delegated auth is a lower-blast-radius execution context than an embedded static credential. Confirm the requested scopes are minimal and that tokens are never logged or forwarded off-host.
An MCP tool surface (manifest or tool definitions) was found.
const server = new Server(
const transport = new StdioServerTransport();
const server = new Server(
server.setRequestHandler(ListToolsRequestSchema, async () => {server.setRequestHandler(CallToolRequestSchema, async (request) => {const transport = new StreamableHTTPServerTransport({const server = new Server(
server.setRequestHandler(ListToolsRequestSchema, async () => {server.setRequestHandler(CallToolRequestSchema, async (req) => {const transport = new StdioServerTransport(process.stdin, process.stdout, { maxBufferSize: 160 * 1024 * 1024 });Why it matters: Just context — review which tools it offers and their permissions.
Fix: Review the declared MCP tools and their permissions.
How attackers abuse these capabilities
Interactive labs on the attack class behind the rules above. They show the technique, not anything found in relay-companion.
Maintain relay-companion?
The free Claude Code plugin scans each npm, PyPI or MCP package with this report's open-source engine before the install command runs. It runs on your machine and needs no account.
Check your own component
Run the same evidence-backed scan on any MCP server, agent skill, or package.
Scan your own componentHow we determine this: deterministic static analysis (regex + AST), evidence-anchored, no code execution. Methodology →