registry / @onescience/onecode / 1.14.50-202607010646

@onescience/onecode@1.14.50-202607010646

OneScience AI coding agent for the terminal.

AI Security Review

scanned 3mo ago · by lpm-firewall-ai

The install hook fetches and installs an executable payload from a hardcoded IP endpoint with TLS verification disabled. The packaged CLI later runs that installed binary.

Static reason

One or more suspicious static signals were detected.

Trigger

npm install runs postinstall; invoking onecode runs the installed binary

Impact

Remote endpoint can supply arbitrary executable code outside npm package integrity controls

Mechanism

install-time unauthenticated remote binary download and execution wrapper

Attack narrative

On install, postinstall.mjs builds a URL under a hardcoded IP host, downloads a tarball with certificate verification disabled, extracts it, moves the extracted package into node_modules, chmods its native binary, and links or copies it to bin/.onecode. The public bin/onecode wrapper resolves that cached binary and spawns it with user arguments.

AI rationale

This package delegates its installed executable to unauthenticated install-time network content, creating a concrete remote code execution supply-chain risk even though the wrapper is CLI-aligned. No source evidence shows credential theft or AI-agent control-surface hijacking.

Evidence

package.jsonpostinstall.mjsbin/onecodenode_modules/onecode-<platform>-<arch>bin/.onecode/tmp/onecode-<platform>-<arch>-<version>.tgz/tmp/onecode-<platform>-<arch>-<version>-extract

Network endpoints1

/onecode_tgz

@onescience/onecode@1.14.50-202607010646

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Source & flagged code