Loading vulnerability…
CVE-2023-20696 is a local privilege-escalation vulnerability in the MediaTek preloader. The issue is caused by a missing bounds check that can lead to an out-of-bounds write during preloader processing. Additional technical reporting indicates the vulnerable code path involves parsing of the boot certificate chain with an ASN.1 routine, where inconsistent interpretation of the certificate container between verification and later parsing can allow attacker-controlled firmware hash data to be accepted as authentic. The flaw affects the early boot chain and can be triggered without user interaction. Successful exploitation can result in execution with System privileges in the boot process context.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
2 valid exploits after Mallory filtered fakes, detection scripts, and README-only repos.
This repository is a Python-based exploit/tooling set for abusing a logic flaw in MediaTek boot-image certificate verification, described as similar to CVE-2023-20696. It is not a remote exploit; it is an offline image-manipulation toolkit used to prepare maliciously modified boot-chain images for vulnerable MTK devices. Repository structure: 7 Python utilities plus README/LICENSE. The core exploit logic is in sign_mtk_cert.py, which parses MTK part_hdr_t images, locates CERT2, extracts existing image-header and image hash fields, recomputes hashes for modified content, and writes a forged ASN.1/DER prefix block containing attacker-controlled OID/hash TLVs. In legacy mode, it additionally wraps the original CERT2 DER in a BIT STRING so older bypass_mode=1 parsers step into attacker-chosen content while still encountering the original certificate structure. This directly matches the README’s exploitation concept for V5/old V6 and new V6 devices. Supporting files: - parse-part-img.py: splits or dumps MTK multi-image containers so components like lk, bl2_ext, lk_main_dtb, and aee can be extracted. - build-part-img.py: replaces or concatenates sub-images back into a multi-image container after patching/signing. - parse_preloader.py: parses MTK preloaders, detects image type/architecture, extracts load addresses, and inspects policy_part_map/reserved memory structures useful for patch development. - parse_mtk_certs.py: pure-Python ASN.1 DER/TLV parser for CERT1/CERT2 blobs with absolute offsets. - verify_mtk_image.py: verifies CERT1/CERT2 RSA-PSS signatures and compares image/header hashes; useful for validation and reverse engineering. - parse_da.py: parses MediaTek Download Agent binaries and detects ARM32/ARM64 parts. Exploit capability: The toolkit enables an operator to modify signed bootloader-stage images while preserving or spoofing certificate/hash data in a way that vulnerable MTK parsers accept. The README states the intended chain is to patch BL2_EXT to stop verifying lk/atf, patch LK to stop verifying lk_main_dtb and remove bootloader locks, then rebuild the image. Successful use yields execution of attacker-controlled boot-chain code and potentially EL3 control. No network communication, C2, hardcoded IPs, or URLs are present in the exploit code. The observable artifacts are local file paths, MTK image magic values, and ASN.1 OIDs used to identify and manipulate certificate fields.
This repository is a Python-based exploit toolkit for abusing a logic flaw in MediaTek boot-image certificate verification, described as similar to CVE-2023-20696. It is not a remote exploit or framework module; it is an offline image-manipulation toolchain intended to modify signed MTK boot components while preserving acceptance by vulnerable preloader/libsec verification code. Repository structure: the codebase contains seven Python utilities plus README/LICENSE. The workflow centers on four main scripts: parse-part-img.py splits MTK multi-image containers using part_hdr_t headers; parse_preloader.py analyzes preloader binaries to recover load addresses, architecture, and policy-related structures; sign_mtk_cert.py is the core exploit component that parses CERT2 ASN.1 data, recalculates image/header hashes, and prepends attacker-controlled DER/TLV structures to exploit parsing behavior; build-part-img.py reinserts the modified single-image back into the original multi-image container. Supporting tools include parse_mtk_certs.py for ASN.1/CERT1/CERT2 inspection, verify_mtk_image.py for local verification of signatures and hashes, and parse_da.py for parsing MediaTek Download Agent binaries. Main exploit capability: sign_mtk_cert.py locates CERT2 in an MTK image, extracts the original image hash and image-header hash OIDs, recalculates hashes over the modified image/header, and writes a new CERT2 blob with prepended override TLVs. For legacy/old devices it can also wrap the original CERT2 DER in a BIT STRING (--legacy), matching the README’s description of bypass_mode=1 behavior where the parser steps into arbitrary objects such as tag 0x03. For newer V6 devices it prepends hash material in a way intended to satisfy bypass_mode=0 parsing while leaving the rest of the certificate structure intact. This is a practical exploit-enablement tool because it automates the certificate manipulation needed to boot altered images. Attack surface and targeting: the attack vector is local/file-based. The operator must possess firmware images and a way to flash or deploy them to a target MTK device. The targeted components are early boot images such as lk, bl2_ext, atf, and related sub-images inside MTK part_hdr_t containers. The README explicitly claims the outcome is gaining EL3 control after modifying image content. No network infrastructure is present. There are no HTTP/DNS/IP callbacks, C2 endpoints, or sockets. Extracted endpoints are limited to local file/script names, MTK image magic strings, and ASN.1 OIDs used to identify certificate fields. Overall, this is a real exploit-support repository focused on boot-chain certificate bypass and signed-image tampering for vulnerable MediaTek devices.
Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.
Vendor-confirmed product mapping. Mallory continuously reconciles this list against your asset inventory.
4 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.