dtale CVE-2026-19896

Exploitation requires that dtale is running with the default build_secret_key function generating its Flask SECRET_KEY, meaning no externally configured SECRET_KEY has been set via environment variable or Flask configuration override. The attacker must be able to capture at least one valid Flask session cookie from the target instance, requiring a network-level position to observe HTTP traffic such as a MITM position, same network segment access, or interception of an internet-exposed dtale endpoint. Single-user or localhost-only dtale deployments where the attacker cannot intercept cookie traffic are effectively protected by inaccessibility. Exploitation is further constrained by high attack complexity (AC:H) per the CVSS 4.0 vector, reflecting the computational and positional prerequisites. No attacker authentication to the dtale application is required prior to the attack (PR:N).

The CVSS 4.0 score of 2.9 with AC:H reflects genuine exploitation difficulty - the attacker must first capture a valid session cookie, then characterize and exploit the NumPy PRNG state to derive the key, which is computationally demanding but theoretically feasible given the constrained key space. Even a mid-sized analytics team in sanok relying on dtale for shared internal data exploration should weigh that the published proof-of-concept meaningfully reduces the practical barrier despite the high complexity rating. EPSS data was not provided in the available intelligence, leaving probabilistic exploitation likelihood unquantified. The limited impact scope - VC:L only in the provided CVSS 4.0 vector, with no integrity or availability impact - tempers urgency significantly; however, the absence of a merged patch means all deployments up to version 3.22.0 remain unmitigated. Risk is most significant in multi-user, network-exposed dtale instances where session cookies are observable; local single-user deployments where traffic interception is not feasible face negligible real-world exposure.

An attacker with access to the same network segment as a multi-user dtale deployment intercepts HTTP traffic to obtain a valid Flask session cookie. Using the published proof-of-concept, the attacker exploits the deterministic NumPy PRNG by exhausting the constrained 10-character, 36-symbol key space to recover the SECRET_KEY. With the recovered key, the attacker crafts a forged Flask session cookie signed with the correct HMAC, impersonating a legitimate user and gaining unauthorized access to that user's dtale session and any datasets visible within it.