opencode-openai-codex-auth
Fail
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONMETADATA_POISONINGCREDENTIALS_UNSAFEDYNAMIC_EXECUTION
Full Analysis
- [METADATA_POISONING]: The skill claims to enable access to 'GPT-5.x' and 'GPT-5.2' models. As these models are not officially available as public products, these claims are highly suspicious and likely represent deceptive social engineering intended to induce users to install unvetted software.
- [EXTERNAL_DOWNLOADS]: Installation instructions require the use of 'npx' to download and execute the 'opencode-openai-codex-auth' package from the public npm registry. This executes code from an unverified source without version pinning or integrity checks.
- [REMOTE_CODE_EXECUTION]: Manual installation involves cloning a repository from a personal GitHub account ('https://github.com/numman-ali/opencode-openai-codex-auth.git'), which involves downloading and potentially executing code from an unvetted third-party source.
- [CREDENTIALS_UNSAFE]: The skill specifically handles and stores 'OPENAI_SESSION_TOKEN' in the user's home directory at '~/.opencode/auth/session.json'. The combination of managing sensitive authentication tokens while executing unvetted remote code poses a severe risk of credential exfiltration.
- [DYNAMIC_EXECUTION]: The use of 'npx' facilitates the dynamic loading and execution of code fetched from a remote registry at runtime.
Recommendations
- AI detected serious security threats
Audit Metadata