@xaccefy/pi-xpi
XPI — offensive security tools for Pi Agent. Casefile tracking, web search, library docs, exploit technique search, todo tracking, and authenticated-session management.
This report shows the evidence for @xaccefy/pi-xpi@0.5.1. 4 of 20 known versions are blocked or flagged. The newest version, 0.9.4, is allowed. See all versions.
AI Security Review
scanned 3mo ago · by lpm-firewall-aiThe source establishes an unconsented install-time mutation of a broad AI-agent control surface and automatic onboarding of third-party code. This meets the firewall block boundary despite no observed credential theft or direct exfiltration.
Decision evidence
public snapshot- `package.json` runs `bash install.sh` during `postinstall`.
- `install.sh` automatically invokes `pi install` for four third-party extensions.
- On install failure it runs unpinned `npm install --no-save` fallbacks.
- Fallbacks append those third-party packages to `~/.pi/agent/settings.json`.
- The script copies package agent definitions into `~/.pi/agent/agents/` without consent.
- No credential harvesting or outbound exfiltration is present in shipped files.
- No remote shell download, eval, destructive command, or binary execution is present.
- `preview.is` is only printed as an API-key signup instruction.
Source & flagged code
3 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 @xaccefy/pi-xpi@0.5.1.
The versions list below shows the verdict for each recorded version.
- Avoid installing @xaccefy/pi-xpi@0.5.1. Remove it from direct dependencies and check your lockfile for transitive copies.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
- If this version ran, investigate the affected machine and build environment. Rotate credentials it could access and rebuild from a trusted environment.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.