IBM Langflow OSS CVE-2026-17621

Exploitation requires a valid low-privilege account on the IBM Langflow OSS instance (PR:L per CVSS). The target must be running IBM Langflow OSS version 1.0.0 through 1.10.2. No other special configuration or feature enablement is described - the traversal is reachable through the standard URL-routing layer of the application. Exploitation is limited by: the requirement for at least low-privilege authentication (anonymous access is insufficient); access to network connectivity to the Langflow OSS web interface; and the files exposed being limited to those readable by the OS process account running Langflow. Files outside the OS process's read permissions cannot be accessed.

The vendor CVSS of 5.4 (Medium) reflects that low privileges are required (PR:L), which meaningfully limits the attack surface relative to unauthenticated traversal flaws. The integrity impact of I:L in the vendor vector is inconsistent with the description, which describes only file reading ('view arbitrary files') - no write or modification capability is mentioned; this metric appears overstated and should be verified with IBM. The confidentiality impact of C:L is likely understated if the traversal is truly unrestricted, as arbitrary file read can expose SSH private keys, database credentials, API tokens, and application secrets stored on the server. No EPSS score is provided in the input; no CISA KEV listing or public exploit code has been identified, suggesting limited active exploitation at this time. SSVC dimensions: exploitation status is none publicly known; the flaw is automatable once credentials are obtained; technical impact is partial (read-only, no availability/integrity demonstrated). Real-world priority is moderate: the authentication requirement reduces automated mass exploitation risk, but any low-privilege Langflow user could pivot to server-side file disclosure.

An attacker with a valid low-privilege Langflow OSS account crafts HTTP requests containing path traversal sequences such as /api/v1/files/../../../../etc/shadow or similar patterns targeting sensitive configuration files. By iterating traversal depth and path guesses, the attacker can exfiltrate server credentials, API keys, or TLS private keys stored on the system, enabling lateral movement or further compromise. No public exploit code has been identified, but the technique is well-understood and trivially reproducible by any practitioner.