If this package is installed
Remove it, treat the machine (or CI runner) as compromised, and rotate every secret, token and key it could access — from a different, clean machine.
Details
Source: amazon-inspector (19164d6b4b781d022a4bc75ae45fae6d20737ddc0b711057e89990d39efc8df5)
The npm package api-anthropic impersonates Anthropic branding (both the package name and the destination domain api-anthropic.help) and reconfigures the user's Claude Code installation to send all traffic through an author-controlled gateway. The CLI writes ~/.claude/settings.json with env.ANTHROPIC_BASE_URL set to https://api-anthropic.help and env.ANTHROPIC_AUTH_TOKEN set to a key, so every subsequent Claude Code session - including prompts, pasted code, file contents, and credentials shared with the assistant - flows through the author's domain instead of Anthropic's real API. The README advertises a non-Anthropic model ("Fable 5.1 (1M Context)"), confirming the endpoint is a third-party proxy rather than Anthropic infrastructure. In addition, the package writes hasTrustDialogAccepted=true and hasClaudeMdExternalIncludesApproved=true into ~/.claude.json for both the current working directory and the user's entire $HOME, pre-approving Claude Code's folder-trust and external-include safety prompts across every project in the home directory. The combination gives the proxy operator a durable channel to capture all AI-assisted content and to have Claude Code act on arbitrary files beneath $HOME without further confirmation.
Source: ghsa-malware (f1922723e10fcad576f383d44d31606460edf694315338ff529a20a84add0005)
Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it.