@rallycry/conveyor-agent@11.0.8
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 12d ago · by lpm-firewall-aiInspected source does not substantiate the scanner’s malicious label: the manifest has no install hook, and the flagged credential and upload behavior is consistent with the agent runtime. The required flagged paths were inspected; source similarity labels alone do not establish malicious behavior.
Decision evidence
public snapshot- package.json defines only `prepublishOnly` as a lifecycle script; it runs `tsup`.
- The boot code copies Claude credentials to a pod-local file and repoints a symlink within the runtime setup.
- The flagged agent-configuration locations handle Claude project paths and a Grimoire submodule that requires a GitHub token; they do not show credential forwarding to an unrelated recipient.
- File upload is exposed through an attachment tool that takes a path from its invocation and resolves relative paths within the configured workspace.
- No inspected source shows install-time execution, credential exfiltration, or an unconsented write to a foreign AI-agent control surface.
Source & flagged code
18 flagged · loading sourceSource appears to send environment or credential material to an external endpoint.
dist/chunk-SKTCDL4K.jsView on unpkg · L136A package entrypoint or install-time lifecycle script reaches a source file with blocking dangerous behavior.
dist/chunk-SKTCDL4K.jsView on unpkg · L136Findings
Affected versions and remediation
This report applies to @rallycry/conveyor-agent@11.0.8.
See all versions of @rallycry/conveyor-agent for other recorded verdicts.
- Review the evidence and your use of @rallycry/conveyor-agent@11.0.8 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.