gittuf's policy can be rolled back to prior valid versions

Description

Summary

An attacker with push access to gittuf's Reference State Log (RSL) can roll back the current policy to any previous policy trusted by the current set of root keys.

Impact

gittuf determines the policy to load by inspecting the RSL. Except for the very first policy (which is automatically trusted given gittuf's TOFU model, or verified against manually specified keys), whenever an RSL entry that points to a new policy is encountered, gittuf validates that this policy is trusted. This is done by checking that the new policy’s root metadata is signed by the required threshold of the current policy's root keys.

Because of this, an attacker with push access to the RSL may create a new entry that references an old policy (that is trusted by the most recent policy's set of root keys), thereby rolling back gittuf's policy to the attacker's chosen state.

Note that the attacker cannot roll back the policy to one that would no longer be trusted by the most recent policy's root keys. As an example, say that Policy A's root keys were for Alice and Bob with a threshold of two, and then Policy B replaced Bob with Carol, with a threshold of two required. An attacker would not be able to roll back the policy to policy A, because it was signed by Alice and Bob, not Alice and Carol.

If you host your repository on a forge such as GitHub, GitLab, Bitbucket, etc., and only certain people are allowed to push to the RSL (e.g. only maintainers of your project have push access to the repository, and all other contributions must be done via pull request), then only those users with push access can perform the attack, or the forge itself.

Fix

gittuf version v0.14.0 introduces a monotonically increasing number to all policy metadata (not to be confused with the metadata version, which dictates the features supported by the metadata). This number is incremented whenever a gittuf client of version v0.14.0 or greater updates the metadata or invokes a patch command to add this field.

During policy verification, gittuf compares the monotonically increasing number of the new policy it encounters in the RSL to the current policy. If any file in the policy (e.g. root of trust, primary policy file, or delegated policy file) has this number, gittuf will check that it increases by one any time the respective file is changed.

Resolution

Upgrade gittuf and Apply Metadata Patch

gittuf strongly recommends users upgrade to the latest version to remediate this vulnerability. Specifically, gittuf version v0.14.0 (and above) are patched with respect to this vulnerability. All users who use gittuf to verify the repository or update its policy metadata must upgrade to prevent this attack.

After a consuming application applies the upgrade, a root of trust user or policy administrator must run:

gittuf trust increment-version

or:

gittuf policy increment-version

to add the monotonically increasing number field to the metadata (see Fix above).

Check for Rollback Attack Attempts

Because this attack requires the attacker to add an entry to the RSL, and because gittuf stores the RSL inside the repository, this attack cannot be performed without leaving evidence behind (in this case, the malicious RSL entry). To check for whether this attack was performed on your repository, run:

gittuf rsl log --ref refs/gittuf/policy

All RSL entries that record new policy states for gittuf should be displayed. There should be one RSL entry every time a legitimate, new policy was applied. Should users find an RSL entry which points to an older policy state, their repository has been compromised.

Credits

gittuf thanks Andrew Nesbitt (@andrew) for finding and responsibly disclosing this vulnerability.

Basic information

Type
reviewed
Severity
medium
Advisory on GitHub
Open advisory ↗
Repository advisory
Open repository advisory ↗
Source code
Browse source ↗
Published (advisory)
2026-05-07 03:34:13 UTC
Updated
2026-05-14 20:54:38 UTC
GitHub reviewed
2026-05-07 03:34:13 UTC
NVD published
2026-05-14

EPSS Score

Score Percentile
0.04% 11.60%

CVSS Scores

Base score Version Severity Vector
6.0 4.0
CVSS:4.0/AV:N/AC:H/AT:P/PR:L/UI:N/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N Click to expand
Attack vector (AV:N)
Could be attacked over the internet or any normal routed network.
Attack complexity (AC:H)
Exploitation depends on constrained or hard-to-reproduce conditions.
Attack requirements (AT:P)
Additional preconditions must be present for exploitation.
Privileges required (PR:L)
Low privileges are required.
User interaction (UI:N)
No user interaction is required.
Vulnerable system confidentiality impact (VC:N)
No confidentiality impact on the vulnerable system.
Vulnerable system integrity impact (VI:H)
High integrity impact on the vulnerable system.
Vulnerable system availability impact (VA:N)
No availability impact on the vulnerable system.
Subsequent system confidentiality impact (SC:N)
No confidentiality impact on subsequent systems.
Subsequent system integrity impact (SI:N)
No integrity impact on subsequent systems.
Subsequent system availability impact (SA:N)
No availability impact on subsequent systems.

Identifiers

CWEs

CWE id Name
CWE-639 Authorization Bypass Through User-Controlled Key

Credits

  • andrew (reporter)

Affected packages (1)

Vulnerable version ranges and first patched releases as published by GitHub.

Ecosystem Package Vulnerable range First patched Vulnerable functions
go github.com/gittuf/gittuf <= 0.13.1 0.14.0

References

cvelogic Threat Intelligence