{"document":{"category":"csaf_vex","csaf_version":"2.0","title":"CVE-2026-50549: Cursor Desktop sandbox escape via symlink and failed path canonicalization","publisher":{"category":"vendor","name":"HarborGuard Database","namespace":"https://database.harborguard.co"},"tracking":{"id":"CVE-2026-50549","status":"final","version":"1","initial_release_date":"2026-06-25T18:47:45.070Z","current_release_date":"2026-06-25T19:02:41.108Z","revision_history":[{"date":"2026-06-25T18:47:45.070Z","number":"1","summary":"Initial machine-readable export from HarborGuard."}]},"distribution":{"tlp":{"label":"WHITE"},"text":"Public CVE data; freely redistributable."},"notes":[{"category":"description","text":"Cursor is a code editor built for programming with AI. Prior to 3.0, Cursor runs agent terminal commands in a sandbox by default. Before a Write, the agent canonicalizes the target path to confirm it stays inside the workspace, but when canonicalization fails it falls back to the original path and writes without approval. A malicious agent can create an in-workspace symlink that points outside the workspace and force canonicalization to fail — either because the target does not exist or because read permission is removed from the path — so the agent writes through the symlink to an arbitrary location without approval. A malicious agent could write arbitrary files outside the workspace under the user's privileges. This enables non-sandboxed Remote Code Execution — for example by overwriting the cursorsandbox helper so later commands run unsandboxed — with no user interaction beyond a benign prompt. This vulnerability is fixed in 3.0.","title":"CVE description"}],"references":[{"category":"self","summary":"CVE-2026-50549 on HarborGuard Database","url":"https://database.harborguard.co/cve/CVE-2026-50549"},{"category":"external","summary":"CVE Record","url":"https://www.cve.org/CVERecord?id=CVE-2026-50549"},{"category":"external","summary":"https://github.com/cursor/cursor/security/advisories/GHSA-3v8f-48vw-3mjx","url":"https://github.com/cursor/cursor/security/advisories/GHSA-3v8f-48vw-3mjx"}]},"product_tree":{"branches":[{"category":"vendor","name":"cursor","branches":[{"category":"product_name","name":"cursor","branches":[{"category":"product_version","name":"< 3.0","product":{"name":"cursor cursor < 3.0","product_id":"CSAFPID-1","product_identification_helper":{"cpe":"cpe:2.3:a:cursor:cursor:\\<_3.0:*:*:*:*:*:*:*"}}}]}]}]},"vulnerabilities":[{"cve":"CVE-2026-50549","title":"Cursor Desktop sandbox escape via symlink and failed path canonicalization","notes":[{"category":"description","text":"Cursor is a code editor built for programming with AI. Prior to 3.0, Cursor runs agent terminal commands in a sandbox by default. Before a Write, the agent canonicalizes the target path to confirm it stays inside the workspace, but when canonicalization fails it falls back to the original path and writes without approval. A malicious agent can create an in-workspace symlink that points outside the workspace and force canonicalization to fail — either because the target does not exist or because read permission is removed from the path — so the agent writes through the symlink to an arbitrary location without approval. A malicious agent could write arbitrary files outside the workspace under the user's privileges. This enables non-sandboxed Remote Code Execution — for example by overwriting the cursorsandbox helper so later commands run unsandboxed — with no user interaction beyond a benign prompt. This vulnerability is fixed in 3.0.","title":"CVE description"}],"product_status":{"known_affected":["CSAFPID-1"]},"scores":[{"cvss_v4":{"version":"4.0","vectorString":"CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N","baseScore":9.3,"baseSeverity":"CRITICAL"},"products":["CSAFPID-1"]}],"remediations":[{"category":"none_available","details":"No fixed version is published yet. Monitor the upstream advisory.","product_ids":["CSAFPID-1"]}]}]}