/

CVE-2026-46699

Severity:

HIGH RISK

(7.6)

Authorization Flaw (CWE-284) in conda-smithy Webservices

CVE-2026-46699

Deferred

Vulnerability type

Category:

Authorization and Access Control

CWE:

CWE-284

Impact:

Unauthorized Repository Write Access and Supply Chain Risk

Severity:

HIGH RISK

(7.6)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

A broken authorization check in the conda-forge automated webservices behind conda-smithy allowed repository access decisions to be based on mutable GitHub usernames instead of immutable user IDs.

If an attacker could take over or re-register a targeted GitHub username, they could gain unintended write access to associated feedstock repositories and tamper with build configurations or packages.

Affected context

This vulnerability occurs in the automation layer that manages access to conda-forge feedstock repositories created or updated by conda-smithy.

It is relevant in environments where:

  • GitHub usernames are used as the primary key for routing repository invitations or access decisions
  • Automated webservices maintain long-lived mappings from usernames to repository permissions
  • Username changes, account deletion, or takeover (e.g., re-registration of abandoned names) are possible

Exploit preconditions

  • Attack surface: The conda-forge automation and GitHub integration that grant or manage feedstock repository access.
  • Required access: Ability to control a GitHub account whose username matches one previously trusted by the system (via takeover, rename, or re-registration scenarios).
  • Trigger point: The automated webservice processes an access or invitation routing decision using the compromised or reused username.
  • Why exploitation is possible: Authorization logic binds permissions to mutable GitHub usernames rather than immutable user IDs, allowing a different principal to inherit prior permissions if they gain control over the name.

Why this happens

The root cause is an authorization design flaw where access control decisions depend on mutable, non-unique identifiers (GitHub usernames) instead of stable, immutable identifiers (GitHub user IDs).

This maps to CWE-284: Improper Access Control, specifically a misbinding of identity and authorization, where the system assumes that a username always corresponds to the same account.

When a username is changed, abandoned, or taken over, the original trust relationship is incorrectly transferred to a new principal, enabling unintended access to feedstock repositories.

Potential impact

  • An attacker may gain unintended write access to conda-forge feedstock repositories.
  • Malicious modifications could be pushed to recipes or build configurations.
  • Attackers could insert backdoors, vulnerable dependencies, or tampered artifacts into published packages.
  • The conda-forge ecosystem and downstream users could be exposed to supply chain compromise.

General mitigation guidance

Mitigation generally focuses on hardening identity and authorization bindings in automation services.

Common defensive patterns include:

  • Using immutable account identifiers (e.g., GitHub user IDs) rather than mutable usernames for permission mapping.
  • Regularly reconciling and validating identity mappings against the identity provider (e.g., GitHub API) before granting access.
  • Implementing explicit ownership and membership verification flows when usernames change.
  • Limiting long-lived, cached authorization decisions and re-checking permissions at key operations.
  • Applying least privilege to automated webservices, ensuring they only grant necessary repository rights.

How teams detect issues like this

Detection strategies typically aim to identify suspicious identity transitions and anomalous repository changes:

  • Reviewing automation code for reliance on mutable identifiers (usernames, emails) in authorization checks.
  • Implementing static analysis or policy checks that flag authorization logic not keyed on immutable IDs.
  • Monitoring audit logs for changes in repository collaborators or permissions associated with username changes.
  • Setting up alerts for unusual commit patterns or configuration changes in feedstock repositories, especially shortly after username updates.
  • Using automated security review tools to scan CI/CD integrations and webhooks for insecure identity assumptions.

Sources

SourceLink
github.comgithub.com/conda-forge/conda-smithy/commit/3b0bcd92ebd6f41edd341401d84583a20911c587
github.comgithub.com/conda-forge/conda-smithy/security/advisories/GHSA-g95q-3cmj-fvh8

Frequently asked

questions

Exploitability depends on the ability to control or take over specific GitHub usernames that the automation previously trusted.

It is not a generic remote exploit; the attacker must align their account identity with a name that already holds or routes access within the system.

The attacker needs control of a GitHub account whose username matches one used by the conda-forge automation for invitation or permission routing.

This may stem from username takeover, re-registration of abandoned names, or gaining control of an existing account.

The most robust fix is to base all authorization and invitation routing on immutable, provider-issued identifiers such as GitHub user IDs, and to ensure that permission checks always resolve against the current identity state from GitHub.

Preventive measures include designing systems to never rely on mutable identifiers for access control, enforcing least privilege in automation, auditing identity and permission mappings regularly, and incorporating security reviews for CI/CD and SCM integration code that manages repository access.

Find out what's already
exploitable in your codebase.