u1s1-cli@1.11.0
u1s1 — 有一说一,最省心的 AI 编程搭子。终端里用中文说需求,AI 帮你读文件、改代码、跑命令。
AI Security Review
scanned 1mo ago · by lpm-firewall-aiThe package does not meet the combined lifecycle self-dependency or concrete malicious-behavior threshold. Its postinstall mutation of potentially hoisted AI-agent dependency files is nevertheless an unresolved, non-consensual lifecycle risk.
Decision evidence
public snapshot- The automatic postinstall hook runs a script that patches another package's installed code.
- The patch locator walks upward through parent directories, so it can modify hoisted copies outside this package's own directory.
- The hook overwrites Pi coding-agent UI and extension-handling files, including suppressing extension stack output unless an environment flag is set.
- Normal CLI startup writes and loads several extensions and a system-prompt append file in its agent directory.
- There is no runtime dependency on u1s1-cli itself.
- The reviewed network and binary-download behavior is invoked during CLI use and is aligned with its agent/search features.
- The code explicitly strips its own secret environment variables before child processes are launched.
- No source evidence shows credential exfiltration, destructive commands, or an opaque remote payload in the install hook.
Source & flagged code
16 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgPackage source references child process execution.
dist/shell-doctor.jsView on unpkg · L1Findings
Affected versions and remediation
This report applies to u1s1-cli@1.11.0.
See all versions of u1s1-cli for other recorded verdicts.
- Review the evidence and your use of u1s1-cli@1.11.0 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.