Local Verification Layer Vulnerability in Mareforma Affects AI-Assisted Research
CVE-2026-71535

7.1HIGH

Key Information:

Vendor

Mareforma

Status
Vendor
CVE Published:
18 November 2026

What is CVE-2026-71535?

The Mareforma platform, used for AI-assisted research, contains a security issue in the local verification layer. Versions 0.3.0 to 0.3.7 feature a flaw in the 'maybe_update_replicated_unlocked' convergence path found in 'mareforma/db/core.py'. This flaw permits the use of generated 'by' labels without validating the associated 'signature_bundle' from an authorized signer. As a result, an unsigned claim can be integrated into a shared graph, allowing it to converge with a signed peer, thus being improperly promoted to a REPLICATED state. This alteration facilitates the return of a forged trust status through queries dependent on REPLICATED support, potentially misleading downstream consumers with false endorsements from multiple parties. The vulnerability has been addressed in version 0.3.7.

Affected Version(s)

mareforma >= 0.3.0, < 0.3.7

References

CVSS V3.1

Score:
7.1
Severity:
HIGH
Confidentiality:
None
Integrity:
High
Availability:
None
Attack Vector:
Local
Attack Complexity:
Low
Privileges Required:
None
User Interaction:
None
Scope:
Changed

Timeline

  • Vulnerability published

  • Vulnerability Reserved

.