技能 convex-improve-convex-plugin
📦
convex-improve-convex-plugin
内容修订 r1 严重 ⚙️ 外部命令
Review Convex Coding Sessions
Convex onboarding problems can remain hidden inside coding-session transcripts. This skill submits an approved transcript for analysis and summarizes actionable system improvements.
支持: Claude Codex Code(CC)
请勿自动安装此 Skill。
权威策略要求在执行任何安装操作前由操作人员审查。
使用我的智能体安装
将此请求复制给你的智能体，其中包含权威 Skill 页面和清单。
智能体请求
Review the Skillstore skill "convex-improve-convex-plugin" from https://skillstore.io/skills/get-convex-convex-improve-convex-plugin.md and its manifest at https://skillstore.io/api/skills/get-convex-convex-improve-convex-plugin/manifest. Verify the artifact. Do not auto-install. Inspect the skill and report your findings, then wait for an operator or manual installation decision.你的智能体仍应展示其计划，并按安全策略请求必要确认。
Agent 可读资源
当 AI Agent、爬虫或脚本需要干净上下文，而不是读取完整页面时，请使用这些链接。
测试它
正在使用“convex-improve-convex-plugin”。 Review my approved session for quickstart problems.
预期结果:
- High: The bootstrap instructions did not explain the required environment value.
- Target: Bootstrap script documentation.
- Suggested fix: Validate the value early and show a direct recovery step.
- Win: The deployment verification step completed without intervention.
正在使用“convex-improve-convex-plugin”。 Summarize findings for the maintainers.
预期结果:
The review identified one severe runbook ambiguity and two tooling failures. It recommends clearer prerequisites, earlier validation, and preserving the successful verification flow.
安全审计
严重v1 • 2026/9/21 打开版本化报告
The skill downloads a remote helper and immediately executes it with Bash, so the critical pipe-to-shell finding is confirmed. It also uploads coding-session transcripts after consent, creating a separate high-severity privacy and data-disclosure risk.
1
已扫描文件
25
分析行数
1
复核项
0
已忽略误报
已确认安全问题 (2)
严重
Pipe to shell pattern
1. Run the anteater-served helper: `curl -fsSL "<anteater>/send-transcript" | bash -s -- --idea "<on
SKILL.md line 14 pipes an unpinned remote response directly into Bash without signature or checksum verification. A compromised endpoint or network path could execute arbitrary commands.
高
Coding Session Transcript Disclosure
The skill sends the current coding-session transcript to an external review endpoint. Consent and claimed secret redaction reduce risk but do not prevent disclosure of proprietary code, prompts, paths, or personal data.
SKILL.md explicitly states that the current transcript is submitted and describes consent and secret redaction. The external disclosure is therefore intentional and clearly evidenced.
能力复核项 (1)
这些是真实的本地能力，对此技能可能是预期行为，因此需要复核，但不会按已确认恶意行为计分。
中
Ruby/shell backtick execution
1. Run the anteater-served helper: `curl -fsSL "<anteater>/send-transcript" | bash -s -- --idea "<on
SKILL.md line 14 instructs the agent to invoke curl and Bash. This is genuine external command execution, even though the static pattern label references Ruby backticks.
风险因素
⚙️ 外部命令 (1)
检测到的模式
Pipe to shell pattern
审计者: codex
分享与引用此报告
分享版本化评估报告、中性徽章、嵌入卡片和引用。Skillstore 只报告证据，不判断该 Skill 是否安全。
复制报告链接
https://skillstore.io/skills/get-convex-convex-improve-convex-plugin/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/get-convex-convex-improve-convex-plugin?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/get-convex-convex-improve-convex-plugin?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/get-convex-convex-improve-convex-plugin/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/get-convex-convex-improve-convex-plugin.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>学术引用 (APA · BibTeX · CFF)
APA 引用
get-convex. (2026). convex-improve-convex-plugin security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/get-convex-convex-improve-convex-plugin/audits/1BibTeX 引用
@techreport{get-convex-get-convex-convex-improve-convex-plugin-2026,
author = {get-convex},
title = {convex-improve-convex-plugin security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/get-convex-convex-improve-convex-plugin/audits/1},
note = {Author version unspecified}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "convex-improve-convex-plugin security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
- name: "get-convex"
date-released: "2026-09-21"
url: "https://skillstore.io/skills/get-convex-convex-improve-convex-plugin/audits/1"
identifiers:
- type: other
value: "skillstore:get-convex-convex-improve-convex-plugin:audit:1"
description: "Skillstore immutable audit report identifier"
Skillstore 评分
评分依据 证据置信度: 低 55
架构
85
可维护性
87
内容
65
社区
83
规范符合性
你能构建什么
Diagnose onboarding friction
Review a completed Convex setup session and identify unclear instructions, stalled steps, and tooling failures.
Improve agent instructions
Extract recurring agent difficulties and suggested fixes for skills, runbooks, and bootstrap scripts.
Prioritize developer experience work
Summarize severe workflow problems and successful patterns from an approved session review.
试试这些提示
Review this session
Review this Convex coding session. Ask for my sharing choice before submitting any transcript.
Find setup friction
Review this session for ambiguous setup instructions, stalled agent behavior, and tooling failures. Summarize the most severe findings and suggested fixes.
Assess a quickstart workflow
Submit this approved session with the application idea: [IDEA]. Group findings by runbook, bootstrap script, skill, or component.
Produce a prioritized improvement brief
Review this approved Convex session. Rank findings by severity, name each affected target, propose concrete fixes, and conclude with successful patterns worth preserving.
最佳实践
- Review the sharing destination and transcript scope before granting consent.
- Use a concise application idea that accurately describes the reviewed session.
- Present findings by severity, affected target, suggested fix, and successful pattern.
避免
- Do not submit a transcript before the user explicitly chooses a consent option.
- Do not assume secret redaction removes proprietary code, personal data, or sensitive context.
- Do not present system-improvement findings as end-user product feedback.
常见问题
What does this skill review?
It reviews a supported coding-session transcript for problems and successes in Convex quickstarts, runbooks, scripts, skills, and components.
Does it upload my transcript?
Yes. After consent, the helper submits the current transcript to an external review endpoint.
Will it ask before sharing?
The workflow requires an explicit Always, Just this once, or Never choice when no saved preference exists.
Are secrets removed?
The skill states that the helper redacts keys and tokens. You should still inspect the data scope before submission.
What happens when no transcript is found?
The helper reports REVIEW_NO_TRANSCRIPT, and the skill informs the user that no Claude or Codex transcript was available.
What results does the skill provide?
It summarizes severe findings with affected targets and suggested fixes, followed by successful patterns from the session.
开发者详情
作者
get-convex许可证
MIT
Skillstore 修订
r1
版本提示
作者未声明版本。
引用
ee1e1d0465cbfd49d2dffacf50000ec9a8812bf0
维护新鲜度
2026/9/21
使用量
0 次下载 · 0 次浏览
文件结构
📄 SKILL.md