@markjaquith/agency@3.3.1
Manage agentic work across repositories with durable workbases
AI Security Review
scanned 1mo ago · by lpm-firewall-aiThe package automatically modifies a global AI-agent extension surface during postinstall. This is a real lifecycle risk but does not meet the policy threshold for a malicious publish block.
Decision evidence
public snapshot- Its postinstall hook automatically installs a Pi agent extension.
- The installer creates and copies code into the user's Pi extension directory.
- The installed extension adds Agency and workbase instructions to an agent system prompt.
- No runtime dependency on this package itself is declared.
- No credential collection, HTTP client, or data exfiltration path was found in the inspected lifecycle and extension code.
- The extension only activates after it finds and validates an Agency workbase.
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@3.3.1.
See all versions of @markjaquith/agency for other recorded verdicts.
- Review the evidence and your use of @markjaquith/agency@3.3.1 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.