agentscope-ai / agentscope-ai/AgentTeams

Security review: 3 validated findings from ClawGuard scan

未關閉
#626 3 則留言 0 個 reaction 已指派 1 人 已被 @maplefeng-a 認領 在 GitHub 檢視
area:worker-runtime
主要語言
Go
星號
5.6k
分支
692
平均合併
5 天 4 小時
30 天內合併 PR
23

描述

Hi, I ran a ClawGuard scan on this repo and manually validated a few concrete findings that look worth tightening:

1. `README.md:49`
- The quick-start docs advertise `curl | bash` as the setup path.
- Exact text: `One Command Setup: curl | bash and you're done`
- This normalizes remote-script execution without pinning or integrity verification.

2. `copaw/src/copaw_worker/sync.py:448`
- The sync cleanup path removes directories with `shutil.rmtree(child)`.
- In context this is deleting local skill directories based on remote state, which deserves extra path safety checks and guardrails before recursive delete.

3. `copaw/src/copaw_worker/worker.py:466`
- Another cleanup path removes directories with `shutil.rmtree(child)` when skills are considered stale.
- Same concern here: recursive delete on dynamically selected paths should be tightly constrained and validated.

I skipped several scanner hits that looked like likely false positives and only kept the ones above because they were easy to confirm directly.

If useful, I can share the full scan output in a follow-up.

Scanner: https://github.com/hesoyam2221/clawguard

貢獻指南

這個儲存庫沒有索引到貢獻指南

評估

這個 Issue 還沒有評估資料。

把新 issue 寄到你的電子郵件信箱

精選適合新手參與的 GitHub issue 摘要。