{"id":"CVE-2026-76894","aliases":["EUVD-2026-116490","GHSA-vqff-r82h-9crc"],"status":"published","title":"CKAN MCP Server: DataStore Table UI resource does not validate postMessage origin (in) and posts to targetOrigin \"*\" (out)","description":"CKAN MCP Server is a tool for querying CKAN open data portals. Prior to 0.4.112, the DataStore table UI in src/resources/datastore-table-ui.ts accepts window message events without validating event.origin and sends JSON-RPC messages through postMessage with a wildcard target origin. A frame that can reach the embedded document can inject fabricated fields or records, spoof pending ui/initialize responses, disrupt the MCP UI handshake, or receive outbound messages when the resource is embedded by an untrusted parent. Values are HTML escaped, so the demonstrated content injection is display spoofing rather than script execution, and exploitability depends on the host embedding model. This issue is fixed in version 0.4.112.","severity":{"version":"3.1","score":3.1,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:L/I:N/A:N","level":"low","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-07","score":0.00196,"percentile":0.09416,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":false,"fix_available":true},"cwes":["CWE-345","CWE-940"],"published":"2026-10-16T19:23:48Z","modified":"2026-11-04T10:47:24Z","fixes":[{"ecosystem":"npm","package":"@aborruso/ckan-mcp-server","fixed_in":["0.4.112"],"sources":["ghsa","osv"]}],"affected":[{"source":"ghsa","ecosystem":"npm","package":"@aborruso/ckan-mcp-server","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"fixed":"0.4.112"}]}]},{"source":"osv","ecosystem":"npm","package":"@aborruso/ckan-mcp-server","ranges":[{"type":"SEMVER","events":[{"introduced":"0"},{"fixed":"0.4.112"}]}]}],"products":[{"vendor":"ondata","product":"ckan-mcp-server"}],"references":[{"url":"https://github.com/ondata/ckan-mcp-server/security/advisories/GHSA-vqff-r82h-9crc","tags":["x_refsource_CONFIRM","advisory","web"]},{"url":"https://github.com/ondata/ckan-mcp-server/commit/3b827af72b228d42aa7b0d7dac5347d16af3f4e5","tags":["x_refsource_MISC","fix"]},{"url":"https://github.com/ondata/ckan-mcp-server/releases/tag/v0.4.112","tags":["x_refsource_MISC","web"]},{"url":"https://github.com/advisories/GHSA-vqff-r82h-9crc"},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/76xxx/CVE-2026-76894.json","tags":["advisory"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-76894","tags":["advisory"]},{"url":"https://github.com/ondata/ckan-mcp-server","tags":["package"]}],"sources":{"cve":[{"id":"CVE-2026-76894","title":"CKAN MCP Server: DataStore Table UI resource does not validate postMessage origin (in) and posts to targetOrigin \"*\" (out)","description":"CKAN MCP Server is a tool for querying CKAN open data portals. Prior to 0.4.112, the DataStore table UI in src/resources/datastore-table-ui.ts accepts window message events without validating event.origin and sends JSON-RPC messages through postMessage with a wildcard target origin. A frame that can reach the embedded document can inject fabricated fields or records, spoof pending ui/initialize responses, disrupt the MCP UI handshake, or receive outbound messages when the resource is embedded by an untrusted parent. Values are HTML escaped, so the demonstrated content injection is display spoofing rather than script execution, and exploitability depends on the host embedding model. This issue is fixed in version 0.4.112.","status":"published","published":"2026-11-03T20:33:35.548Z","modified":"2026-11-03T20:33:35.548Z","cvss":[{"version":"3.1","score":3.1,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:L/I:N/A:N","level":"low","assessor":"cna"}],"cwes":["CWE-345","CWE-940"],"updated_at":"2026-11-03T22:58:38.387176Z"}],"euvd":[{"id":"EUVD-2026-116490","description":"CKAN MCP Server is a tool for querying CKAN open data portals. Prior to 0.4.112, the DataStore table UI in src/resources/datastore-table-ui.ts accepts window message events without validating event.origin and sends JSON-RPC messages through postMessage with a wildcard target origin. A frame that can reach the embedded document can inject fabricated fields or records, spoof pending ui/initialize responses, disrupt the MCP UI handshake, or receive outbound messages when the resource is embedded by an untrusted parent. Values are HTML escaped, so the demonstrated content injection is display spoofing rather than script execution, and exploitability depends on the host embedding model. This issue is fixed in version 0.4.112.","status":"published","published":"2026-11-03T20:33:35Z","modified":"2026-11-03T20:33:35Z","cvss":[{"version":"3.1","score":3.1,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:L/I:N/A:N","level":"low"}],"updated_at":"2026-11-03T23:01:39.342753Z"}],"ghsa":[{"id":"GHSA-vqff-r82h-9crc","title":"ckan-mcp-server: DataStore Table UI resource does not validate postMessage origin (in) and posts to targetOrigin \"*\" (out)","description":"## Summary\n\nThe MCP Apps UI resource (an interactive DataStore table rendered in the host's webview/iframe) communicates with its parent via `postMessage`. Its inbound `message` listener performs no `event.origin` validation, and its outbound messages are sent with target origin `'*'`. Any window/frame able to reach the document can inject data into the table and interfere with the MCP UI handshake, and the UI's own JSON-RPC messages are broadcast without an origin restriction.\n\n## Affected code\n\n`src/resources/datastore-table-ui.ts` (inline HTML):\n\n```js\nwindow.addEventListener('message', function (event) {\n  var msg = event.data;                    // no check of event.origin\n  if (!msg || typeof msg !== 'object') return;\n  if (msg.id !== undefined \u0026\u0026 _pending[msg.id]) { ... resolve/reject pending MCP request ... }\n  ...\n  if (data \u0026\u0026 (data.fields || data.records)) initTable(data);   // renders attacker-supplied data\n});\n\n// outbound:\nwindow.parent.postMessage({ jsonrpc: '2.0', id: id, method: method, params: params || {} }, '*');\nmcpNot: window.parent.postMessage({ jsonrpc:'2.0', method:method, params:params||{} }, '*');\n```\n\nThere is no allowlist of accepted origins for inbound messages and no fixed target origin for outbound messages.\n\n## Impact\n\n- **Content spoofing.** A frame that can `postMessage` to the UI can supply\n  arbitrary `fields`/`records`, causing the table to render fabricated rows and\n  links (e.g. a phishing `__link_url`) that appear to come from the queried\n  portal. (Values are HTML-escaped, so this is display spoofing, not script\n  injection — no XSS.)\n- **Handshake/UX interference.** Because inbound messages matching a pending `id`\n  resolve/reject the UI's MCP requests, an attacker frame can spoof responses to\n  the `ui/initialize` handshake or otherwise disrupt the panel.\n- **Outbound leakage.** Posting to `'*'` means the UI's JSON-RPC messages are\n  delivered to whatever the parent origin is; if the resource is ever embedded by\n  a context other than the trusted host, those messages leak to it.\n\nExploitability depends on the host embedding model (in a correct MCP Apps host the\nparent is the trusted host, which limits who can message the frame), so this is a\nhardening / defense-in-depth issue rather than a directly remotely-triggerable\none — hence Low.\n\n## Proof of concept\n\n- `poc/verify-no-origin-check.mjs` — static confirmation that the listener has no\n  origin check and outbound uses `'*'`:\n\n  ```\n  [x] validates event.origin             : NO\n  [x] outgoing postMessage targetOrigin  : '*' (any)\n  RESULT: VULNERABLE\n  ```\n\n- `poc/parent-spoof-poc.html` — a page that embeds the UI and injects fabricated\n  rows cross-origin with no handshake; the UI renders them.\n\n## Remediation\n\n- Validate `event.origin` against the expected host origin on every inbound\n  message; drop messages from other origins.\n- Send outbound `postMessage` with an explicit target origin (the host origin\n  established at initialize time) instead of `'*'`.\n- Optionally, bind to a nonce/channel established during `ui/initialize` so only\n  the negotiated host can drive the panel.","status":"published","published":"2026-10-16T19:23:48Z","modified":"2026-10-16T19:24:46Z","cvss":[{"version":"3.1","score":3.1,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:L/I:N/A:N","level":"low"}],"cwes":["CWE-345","CWE-940"],"updated_at":"2026-11-02T14:41:32.483482Z"}],"nvd":[{"id":"CVE-2026-76894","description":"CKAN MCP Server is a tool for querying CKAN open data portals. Prior to 0.4.112, the DataStore table UI in src/resources/datastore-table-ui.ts accepts window message events without validating event.origin and sends JSON-RPC messages through postMessage with a wildcard target origin. A frame that can reach the embedded document can inject fabricated fields or records, spoof pending ui/initialize responses, disrupt the MCP UI handshake, or receive outbound messages when the resource is embedded by an untrusted parent. Values are HTML escaped, so the demonstrated content injection is display spoofing rather than script execution, and exploitability depends on the host embedding model. This issue is fixed in version 0.4.112.","status":"published","published":"2026-11-03T21:17:43.24Z","modified":"2026-11-03T21:17:43.24Z","cvss":[{"version":"3.1","score":3.1,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:L/I:N/A:N","level":"low","assessor":"secondary"}],"cwes":["CWE-345","CWE-940"],"updated_at":"2026-11-04T00:48:10.212283Z"}],"osv":[{"id":"CVE-2026-76894","title":"CKAN MCP Server: DataStore Table UI resource does not validate postMessage origin (in) and posts to targetOrigin \"*\" (out)","description":"CKAN MCP Server is a tool for querying CKAN open data portals. Prior to 0.4.112, the DataStore table UI in src/resources/datastore-table-ui.ts accepts window message events without validating event.origin and sends JSON-RPC messages through postMessage with a wildcard target origin. A frame that can reach the embedded document can inject fabricated fields or records, spoof pending ui/initialize responses, disrupt the MCP UI handshake, or receive outbound messages when the resource is embedded by an untrusted parent. Values are HTML escaped, so the demonstrated content injection is display spoofing rather than script execution, and exploitability depends on the host embedding model. This issue is fixed in version 0.4.112.","status":"published","published":"2026-11-03T20:33:35.548Z","modified":"2026-11-05T03:31:29.268462463Z","cvss":[{"version":"3.1","score":3.1,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:L/I:N/A:N","level":"low"}],"cwes":["CWE-345","CWE-940"],"updated_at":"2026-11-05T07:12:28.867005Z"},{"id":"GHSA-vqff-r82h-9crc","title":"ckan-mcp-server: DataStore Table UI resource does not validate postMessage origin (in) and posts to targetOrigin \"*\" (out)","description":"## Summary\n\nThe MCP Apps UI resource (an interactive DataStore table rendered in the host's webview/iframe) communicates with its parent via `postMessage`. Its inbound `message` listener performs no `event.origin` validation, and its outbound messages are sent with target origin `'*'`. Any window/frame able to reach the document can inject data into the table and interfere with the MCP UI handshake, and the UI's own JSON-RPC messages are broadcast without an origin restriction.\n\n## Affected code\n\n`src/resources/datastore-table-ui.ts` (inline HTML):\n\n```js\nwindow.addEventListener('message', function (event) {\n  var msg = event.data;                    // no check of event.origin\n  if (!msg || typeof msg !== 'object') return;\n  if (msg.id !== undefined \u0026\u0026 _pending[msg.id]) { ... resolve/reject pending MCP request ... }\n  ...\n  if (data \u0026\u0026 (data.fields || data.records)) initTable(data);   // renders attacker-supplied data\n});\n\n// outbound:\nwindow.parent.postMessage({ jsonrpc: '2.0', id: id, method: method, params: params || {} }, '*');\nmcpNot: window.parent.postMessage({ jsonrpc:'2.0', method:method, params:params||{} }, '*');\n```\n\nThere is no allowlist of accepted origins for inbound messages and no fixed target origin for outbound messages.\n\n## Impact\n\n- **Content spoofing.** A frame that can `postMessage` to the UI can supply\n  arbitrary `fields`/`records`, causing the table to render fabricated rows and\n  links (e.g. a phishing `__link_url`) that appear to come from the queried\n  portal. (Values are HTML-escaped, so this is display spoofing, not script\n  injection — no XSS.)\n- **Handshake/UX interference.** Because inbound messages matching a pending `id`\n  resolve/reject the UI's MCP requests, an attacker frame can spoof responses to\n  the `ui/initialize` handshake or otherwise disrupt the panel.\n- **Outbound leakage.** Posting to `'*'` means the UI's JSON-RPC messages are\n  delivered to whatever the parent origin is; if the resource is ever embedded by\n  a context other than the trusted host, those messages leak to it.\n\nExploitability depends on the host embedding model (in a correct MCP Apps host the\nparent is the trusted host, which limits who can message the frame), so this is a\nhardening / defense-in-depth issue rather than a directly remotely-triggerable\none — hence Low.\n\n## Proof of concept\n\n- `poc/verify-no-origin-check.mjs` — static confirmation that the listener has no\n  origin check and outbound uses `'*'`:\n\n  ```\n  [x] validates event.origin             : NO\n  [x] outgoing postMessage targetOrigin  : '*' (any)\n  RESULT: VULNERABLE\n  ```\n\n- `poc/parent-spoof-poc.html` — a page that embeds the UI and injects fabricated\n  rows cross-origin with no handshake; the UI renders them.\n\n## Remediation\n\n- Validate `event.origin` against the expected host origin on every inbound\n  message; drop messages from other origins.\n- Send outbound `postMessage` with an explicit target origin (the host origin\n  established at initialize time) instead of `'*'`.\n- Optionally, bind to a nonce/channel established during `ui/initialize` so only\n  the negotiated host can drive the panel.","status":"published","published":"2026-10-16T19:23:48Z","modified":"2026-10-16T19:30:03.948336425Z","cvss":[{"version":"3.1","score":3.1,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:L/I:N/A:N","level":"low"}],"cwes":["CWE-345","CWE-940"],"updated_at":"2026-11-02T15:46:57.694388Z"}]}}