@xzy-ai/pi-c2@0.84.2-1.821.2026
PI extension: delegate work to isolated in-process subagents through model-callable tools
AI Security Review
scanned 13d ago · by lpm-firewall-aiThe postinstall hook performs host SDK modification without a consent prompt, creating a concrete guarded agent-extension lifecycle risk. The inspected code applies the bundled patch to the named host SDK, so the evidence does not support a malicious block.
Decision evidence
public snapshot- The postinstall hook runs a script that locates the host pi-coding-agent installation.
- The script applies a bundled patch and writes changed files into that host installation without a consent check.
- The extension scans project agent directories and reads their configuration files.
- The patch targets the named host SDK used by this extension; inspected behavior does not show a foreign or broad agent-control-surface mutation.
- The Exa API key is used in a request to the configured Exa service, with no evidence of an unrelated credential recipient.
Source & flagged code
7 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgFindings
Affected versions and remediation
This report applies to @xzy-ai/pi-c2@0.84.2-1.821.2026.
See all versions of @xzy-ai/pi-c2 for other recorded verdicts.
- Review the evidence and your use of @xzy-ai/pi-c2@0.84.2-1.821.2026 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.