# CVE-2026-72703: Rocq Prover guard checker fixpoint parameter handling flaw

Source: https://junglewise.ai/threats/cve-2026-72703-rocq-prover-guard-checker-fixpoint-parameter-handling-flaw  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-08-24. Updated: 2026-09-25  
License: CC BY 4.0 with a link back

CVE-2026-72703 · Severity: medium · CVSS 6.3 · Published 2026-08-24

Rocq Prover, an interactive theorem prover used for formal verification and mathematical proof development, contains a flaw in its guard checker that incorrectly validates recursive function definitions. An attacker can craft a non-terminating function definition that the prover accepts as valid, allowing them to prove false statements and derive contradictions that undermine the soundness of the proof system.

Technologies: [Rocq Prover](https://junglewise.ai/threats/technologies/rocq-prover). Vendors: [Rocq](https://junglewise.ai/threats/vendors/rocq).

## Quick answers

**Is CVE-2026-72703 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-72703 being exploited in the wild.

**How severe is CVE-2026-72703?** Medium severity, with a CVSS score of 6.3, published 24 August 2026.

**Which products and versions are affected?** Rocq Prover 8.20 through 9.1.x.

**Is there a fix for CVE-2026-72703?** A patch was published on 2026: Fixed in Rocq 9.2.0 Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

**How is CVE-2026-72703 attacked?** It is attacked over the network.

**Am I affected by CVE-2026-72703?** Check whether you run Rocq Prover at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

## Technical details

The vulnerability is a logical inconsistency in the guard checker mechanism (kernel/inductive.ml) that validates recursive function definitions. The find_uniform_parameters function inspects only self-recursive calls within nested mutual fixpoints and incorrectly marks parameters as uniform when no body calls itself, failing to examine cross-calls between different fixpoint bodies. When a parameter grows through such cross-calls, it retains an inherited subterm specification that allows recursive calls to be accepted even though their arguments are not structurally smaller. This permits non-terminating definitions to be admitted as structurally decreasing, enabling construction of a term whose value equals its own successor, yielding a proof of False without requiring axioms or unsafe flags. The vulnerability was introduced in Coq 8.20 and is patched in Rocq 9.2.0.

## Affected products

- Rocq Rocq Prover 8.20 through 9.1.x

## Timeline

- 2026-08-24: disclosed
- 2026: patched: Fixed in Rocq 9.2.0

## References

- [https://github.com/endrazine/rocq-cve-poc-21682](https://github.com/endrazine/rocq-cve-poc-21682)
- [https://github.com/rocq-prover/rocq](https://github.com/rocq-prover/rocq)

## Related threats

- [Rocq Prover kernel soundness bug in universe checking state desynchronization](https://junglewise.ai/threats/cve-2026-72714-rocq-prover-kernel-soundness-bug-in-universe-checking-state)
- [Rocq Prover guard checker higher-order fixpoint bypass](https://junglewise.ai/threats/cve-2026-72705-rocq-prover-guard-checker-higher-order-fixpoint-bypass)
- [Rocq Print Assumptions logic error in universe checking](https://junglewise.ai/threats/cve-2020-37268-rocq-print-assumptions-logic-error-in-universe-checking)