oh-my-codex@0.21.4
Multi-agent orchestration layer for OpenAI Codex CLI
AI Security Review
scanned 2mo ago · by lpm-firewall-aiThe package has a concrete automatic remote-native-code staging path with no independent signature or pinned manifest trust root. Its scope is package-aligned and global-install-only, so the source does not establish malicious intent or justify a publish block.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
Global npm installation followed by use of runtime-backed OMX features.
Impact
A compromised or replaced release manifest can stage native code for later execution; the archive hash does not provide an independent trust anchor.
Mechanism
Automatic remote native-binary hydration followed by cached-binary execution.
AI rationale
The package has a concrete automatic remote-native-code staging path with no independent signature or pinned manifest trust root. Its scope is package-aligned and global-install-only, so the source does not establish malicious intent or justify a publish block.
Evidence
package.jsondist/scripts/postinstall.jsdist/cli/native-assets.jsdist/runtime/bridge.js~/.cache/oh-my-codex/native/
Network endpoints1
github.com
Decision evidence
public snapshotAI called this Suspicious at 88.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The postinstall hook runs automatically and imports the package postinstall module.
- On global installation, postinstall automatically hydrates a native runtime.
- Hydration fetches a release manifest and downloads a native archive whose hash is supplied by that same remote manifest.
- Runtime code selects the cached native binary and executes it later.
Evidence against
- The lifecycle code exits without action for local installs.
- The download target is derived from the package's declared GitHub repository and version.
- Setup writes to Codex configuration only through the explicit omx setup command; no install-time agent configuration mutation was found.
- No credential harvesting or unrelated data exfiltration was found in the inspected paths.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNativeBindingsNetworkShell
Source & flagged code
10 flaggedpackage.jsonView file
•scripts.postinstall = node -e "const fs=require('fs');const p='./dist/scripts/postinstall.js';if(fs.existsSync(p))import(p).then(m=>m.main?.()).catch(e=>console.warn('[omx] Postinstall skipped after a n...
Critical
Red Install Lifecycle Script
Install-time lifecycle script matches a deterministic static-gate block pattern.
package.jsonView on unpkg•scripts.postinstall = node -e "const fs=require('fs');const p='./dist/scripts/postinstall.js';if(fs.existsSync(p))import(p).then(m=>m.main?.()).catch(e=>console.warn('[omx] Postinstall skipped after a n...
High
Install Time Lifecycle Scripts
Package defines install-time lifecycle scripts.
package.jsonView on unpkgdist/autoresearch/runtime.jsView file
1import { execFileSync, spawnSync } from 'child_process';
L2: import { existsSync } from 'fs';
High
Findings
1 Critical6 High6 Medium5 Low
CriticalRed Install Lifecycle Scriptpackage.json
HighInstall Time Lifecycle Scriptspackage.json
HighChild Processdist/autoresearch/runtime.js
HighShelldist/autoresearch/runtime.js
HighTrigger Reachable External Ai Agent Control Surface Mutationdist/cli/setup.js
HighRuntime Package Installdist/mcp/code-intel-server.js
HighSemantic Analysis Limiteddist/cli/api.js#virtual:normalized:round1
MediumDynamic Requiredist/runtime/bridge.js
MediumNetwork
MediumEnvironment Vars
MediumInstall Persistencedist/scripts/smoke-packed-install.js
MediumShips Build Helpersrc/scripts/run-autoresearch-showcase.sh
Affected versions and remediation
This report applies to oh-my-codex@0.21.4.
See all versions of oh-my-codex for other recorded verdicts.
- Review the evidence and your use of oh-my-codex@0.21.4 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.