Agent Skills › nightly-labs/openbot › repository-threat-model

repository-threat-model

GitHub

基于代码仓库证据生成应用安全威胁模型，识别资产、边界、滥用路径及缓解措施。仅在明确请求时执行。

marketplace/production-catalog/skills/repository-threat-model/SKILL.md nightly-labs/openbot

Trigger Scenarios

用户明确要求进行威胁建模涉及应用安全架构分析

Install

npx skills add nightly-labs/openbot --skill repository-threat-model -g -y

More Options

Non-standard path

npx skills add https://github.com/nightly-labs/openbot/tree/main/marketplace/production-catalog/skills/repository-threat-model -g -y

Use without installing

npx skills use nightly-labs/openbot@repository-threat-model

指定 Agent (Claude Code)

npx skills add nightly-labs/openbot --skill repository-threat-model -a claude-code -g -y

安装 repo 全部 skill

npx skills add nightly-labs/openbot --all -g -y

预览 repo 内 skill

npx skills add nightly-labs/openbot --list

SKILL.md

Frontmatter

{
    "name": "repository-threat-model",
    "description": "Produce a repository-grounded application security threat model with assets, boundaries, abuse paths, priorities, and mitigations. Use only when threat modeling is explicitly requested."
}

Repository Threat Model

Build the model from repository evidence rather than a generic checklist. Establish the in-scope components, deployment assumptions, entry points, data stores, external integrations, and trust boundaries. Separate runtime behavior from development and CI tooling.

Identify the assets that drive risk and the attacker capabilities that are plausible for the stated environment. Also record meaningful attacker non-capabilities so severity is not overstated.

Describe a small set of concrete abuse paths. For each one, connect an attacker goal to an entry point, crossed boundary, affected asset, existing controls, and resulting impact. Rank it using qualitative likelihood and impact with a short justification.

Ask only for missing context that would materially change scope or ranking. Mark unresolved assumptions explicitly. Recommend mitigations at the component or boundary where they belong, distinguish existing controls from proposed work, and favor specific changes over generic advice.

Finish by checking that every important entry point and trust boundary is represented and that every architectural claim can be traced to repository evidence.