@rallycry/conveyor-agent@11.1.1
Conveyor Agent Runner v10 - PTY harness for the task chat (SDK harness for audit/project-chat). Agent-as-User architecture with BaseService patterns. Works locally too.
AI Security Review
scanned 6d ago · by lpm-firewall-aiSource inspection found no concrete attack behavior. The scanner's lifecycle and agent-configuration hints resolve to a build-only publish hook, pod/session setup, guarded workspace skill handling, and memory-pressure recovery.
Decision evidence
public snapshot- The manifest's only publish lifecycle script runs tsup; it defines no install-time hook.
- Boot copies legacy Claude credentials to a pod-local file, then wires credential and agent directories for the pod.
- Claude transcripts use the configured Claude home and are stored under a project-specific path.
- The Grimoire setup checks for a matching submodule and a token before fetching; bundled skills are materialized in the workspace.
- The Git token helper updates the Git credential and token files for repository access.
- The OOM watchdog checks for memory pressure and an eligible foreign process group before sending SIGKILL.
- Claude hook settings are written in a temporary session directory and passed into the agent spawn.
- The watchdog can kill an eligible foreign process group with SIGKILL under memory pressure, but the inspected code ties this to its OOM recovery function.
Source & flagged code
9 flagged · loading sourceSource writes persistence or remote-access backdoor material.
dist/boot-3FWYGHSN.jsView on unpkg · L36A package entrypoint or install-time lifecycle script reaches a source file with blocking dangerous behavior.
dist/boot-3FWYGHSN.jsView on unpkg · L36Findings
Affected versions and remediation
This report applies to @rallycry/conveyor-agent@11.1.1.
See all versions of @rallycry/conveyor-agent for other recorded verdicts.
- Review the evidence and your use of @rallycry/conveyor-agent@11.1.1 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.