@markjaquith/agency@3.9.5
Manage agentic work across repositories with durable workbases
AI Security Review
scanned 14d ago · by lpm-firewall-aiThe package performs guarded first-party agent extension setup from an npm postinstall hook, which warrants a warning under the agent integration policy. Inspected source did not establish malicious execution, credential exfiltration, or another concrete attack.
Decision evidence
public snapshot- The postinstall hook copies the package-owned extension into the user's global Pi extensions directory.
- The installer creates or updates the extension only when it can establish ownership of the existing file.
- The extension invokes Agency's own context command for discovered workbases; the inspected flow showed no external endpoint or credential forwarding.
- The requested integration code inspects workbase-local agent and OpenCode configuration paths.
- No concrete malicious behavior was identified in the reviewed installer or extension code.
Source & flagged code
10 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 unpkgPackage source references dynamic require/import behavior.
src/utils/interactive-loader.tsView on unpkg · L3Findings
Affected versions and remediation
This report applies to @markjaquith/agency@3.9.5.
See all versions of @markjaquith/agency for other recorded verdicts.
- Review the evidence and your use of @markjaquith/agency@3.9.5 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.