clavue-v1@1.47.0
Clavue v1: lightweight four-slot (main/light/plan/review) multi-model coding CLI on the execution-first engine — overlapped cross-family review, official Clavue cloud preset (api.clavue.com) alongside any Messages-compatible provider, learned context ceil
AI Security Review
scanned 4d ago · by lpm-firewall-aiSource inspection found package-aligned error reporting with common credential redaction, CLI command execution, Ajv serializer code generation, and a filtered sandbox proxy. These behaviors do not establish a malicious attack surface.
Decision evidence
public snapshot- The manifest exposes a CLI entrypoint and does not list a runtime dependency on clavue-v1 itself.
- Incident reporting is disabled by explicit settings or flags; otherwise it sends error records to the package’s Clavue service.
- The reporter redacts common bearer tokens, API keys, passwords, and authorization values before recording text.
- The command runner passes a command and argument list to spawnSync as part of the CLI’s Mao functionality.
- Ajv’s dynamic function creation is part of compiling a serializer from generated source.
- The sandbox HTTP proxy checks each connection with a filter and rejects connections the filter denies.
- Incident reporting is enabled unless settings or environment flags disable it, and records include error details and client metadata.
- The CLI includes command execution and the bundle includes dynamic code generation, both of which require context to assess.
Source & flagged code
12 flagged · loading sourcePackage source references child process execution.
dist/mao-command.jsView on unpkg · L4Findings
Affected versions and remediation
This report applies to clavue-v1@1.47.0.
See all versions of clavue-v1 for other recorded verdicts.
- Review the evidence and your use of clavue-v1@1.47.0 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.