registry / steal-context / 0.1.2

steal-context@0.1.2

Type /steal in one AI coding chat to pull the latest conversation context from the OTHER tool (e.g. Cursor <-> Kilo Code). A thin, model-aware UX layer on top of `continues`.

AI Security Review

scanned 3h ago · by lpm-firewall-ai

LPM treats this as warn-only first-party agent extension lifecycle risk. On global installation, the package adds /steal commands to Cursor and Kilo Code user command directories. Invoking /steal exports the latest project-scoped local AI transcript into the current project.

Static reason

One or more suspicious static signals were detected.

Trigger

npm install -g steal-context; later /steal or steal-context run.

Impact

AI sessions, including stored tool inputs/results, can be copied into .steal/handoff.md and supplied to another agent.

Mechanism

global AI-agent command installation and local transcript handoff

Rationale

This is a guarded, first-party AI-agent extension setup with a high-impact local context-transfer capability. It is not concrete malware, but its install-time global command mutation warrants a warning under policy.

Evidence

package.jsonscripts/postinstall.mjssrc/install.mjssrc/steal.mjssrc/readers/cursor.mjssrc/readers/kilo.mjssrc/continues.mjstemplates/cursor-steal.mdtemplates/kilo-steal.md~/.cursor/commands/steal.md~/.config/kilo/commands/steal.md.steal/handoff.md

steal-context@0.1.2

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code