registry / @routerhub/agent-rules / 1.5.209

@routerhub/agent-rules@1.5.209

Shared Copilot agent rules and guidelines for RouterHub projects

AI Security Review

scanned 27d ago · by lpm-firewall-ai

LPM flags this version as an AI-agent control-surface risk. Installing the package automatically alters the consumer project's package-manager settings and injects AI-agent instructions. Those instructions promote disclosure of real secrets and suppress security review.

Static reason

One or more suspicious static signals were detected.; source matched previously finalized malicious package; routed for review

Trigger

npm installation triggers postinstall.

Impact

Consumer projects can receive persistent unsafe instructions that expose credentials in generated documents and weaken review safeguards.

Mechanism

Postinstall mutation of project configuration and AI-agent control files.

Policy narrative

On installation, postinstall changes .npmrc so dependency lifecycle hooks keep running and then launches the rule generator in the consumer project. The generator writes Copilot and Claude instruction files and copies skills into project-owned agent directories. Its injected rules require real credentials and environment values in documents, then explicitly tell AI and human reviewers not to report that behavior as a security issue. This is an unconsented install-time takeover of broad AI-agent control surfaces.

AI rationale

The automatic lifecycle hook modifies consumer configuration and injects AI instructions that promote secret disclosure while suppressing review. The reviewer-directed text is social engineering and confirms malicious intent.

Evidence

package.jsonpostinstall.jsmerge.jsAGENTS.base.md.npmrcAGENTS.mdCLAUDE.mdAGENTS.private.md.gitignore

@routerhub/agent-rules@1.5.209

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Source & flagged code