registry / @markjaquith/agency / 2.71.25

@markjaquith/agency@2.71.25

Manage agentic work across repositories with durable workbases

AI Security Review

scanned 15d ago · by lpm-firewall-ai

LPM treats this as warn-only first-party agent extension lifecycle risk. Installation copies a first-party Agency extension into Pi's user extension directory. The extension conditionally exposes project skill paths and adds Agency context to Pi agent sessions in Agency workbases; no exfiltration or remote payload path was found.

Static reason

One or more suspicious static signals were detected.

Trigger

npm postinstall, then a Pi agent session opened within an Agency workbase.

Impact

Changes Pi agent behavior for Agency workbases and grants those sessions project-directed context.

Mechanism

global Pi extension installation with conditional agent prompt/resource hooks

Rationale

The automatic global Pi extension write warrants a warning under the lifecycle policy, but the extension is package-aligned and conditionally scoped to Agency workbases. No concrete malicious chain was established.

Evidence

package.jsonscripts/install-pi-extension.tspi-extensions/agency.tssrc/commands/integration.tssrc/services/IntegrationService.ts~/.pi/agent/extensions/agency.ts

Decision evidence

public snapshot

AI called this Suspicious at 91.0% confidence as Dangerous Capability with low false-positive risk.

@markjaquith/agency@2.71.25

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code