技能 mantis-threat-model

📦

mantis-threat-model

Name: mantis-threat-model
Author: google

內容修訂 r1 低風險 ⚙️ 外部命令📁 檔案系統存取

Build Threat Models from Your Mantis Knowledge Base

Architecture notes often leave attacker access and trust boundaries unclear. This skill turns existing Mantis knowledge into a structured threat model with deployment intent and snapshot provenance.

支援: Claude Codex Code(CC)

📊 74 尚可

Maintain this Skill? Claim it 📊 74 尚可

使用我的智慧代理安裝

將此請求複製給你的智慧代理，其中包含權威 Skill 頁面和清單。

智能體請求

Review the Skillstore skill "mantis-threat-model" from https://skillstore.io/skills/google-mantis-threat-model.md and its manifest at https://skillstore.io/api/skills/google-mantis-threat-model/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.

你的智慧代理仍應顯示其計畫，並依安全政策請求必要確認。

Agent 可讀資源

當 AI Agent、爬蟲或腳本需要乾淨脈絡、而不是讀取完整頁面時，請使用這些連結。

Markdown 詳情 GET /skills/google-mantis-threat-model.md 已簽名 manifest GET /api/skills/google-mantis-threat-model/manifest 已簽名 lockfile GET /api/skills/google-mantis-threat-model/lockfile

測試它

正在使用「mantis-threat-model」。 Generate a model from a KB describing a public API, an authentication service, and a customer database.

預期結果:

System overview: The public API accepts client requests and delegates identity checks to the authentication service.
Deployment intent: Production, because the KB describes an externally reachable service.
Trust boundary: Untrusted requests cross into authenticated application operations at the API authorization layer.
Threat actor: An unauthenticated network attacker can reach the API entry point.
High-risk asset: Customer records require confidentiality and integrity protections.

正在使用「mantis-threat-model」。 Rerun the model using the same recorded snapshot and reuse the existing model content.

預期結果:

Snapshot provenance: The model retains the current snapshot identifier.
Freshness warning: The threat model was not reevaluated this pass and was carried forward from the unchanged snapshot.
Archive action: The prior model is copied to the pass archive before replacement.

安全審計

低風險

v1 • 2026/10/4 開啟版本化報告

All 101 static findings are false positives involving Markdown formatting, expected workspace metadata, or locator checks skipped by this KB-only stage. No evidence found of credential theft, exfiltration, malicious execution, or audit-directed prompt injection. One semantic finding identifies potential archive-history loss when the same pass is rerun.

已掃描檔案

252

分析行數

複核項目

已忽略的誤報

已確認的安全問題 (1)

低

SKILL.md:143-151

Repeated Passes Can Replace Archive History

The same pass number selects the same archive filename on repeated runs. Copying the current model there can replace earlier history without a collision check.

The procedure explicitly copies to a pass-number filename but provides no protection for an existing archive. This establishes a potential history-loss risk, not observed exploitation.

風險因素

⚙️ 外部命令 (50)

📁 檔案系統存取 (7)

SKILL.md:37 SKILL.md:71 SKILL.md:77 SKILL.md:95 SKILL.md:97 SKILL.md:143 SKILL.md:92

審計單位: codex

分享與引用此報告

分享版本化評估報告、中性徽章、嵌入卡片與引用。Skillstore 只報告證據，不判斷該 Skill 是否安全。

開啟版本化報告

複製報告連結

https://skillstore.io/skills/google-mantis-threat-model/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown 徽章

[![Skillstore security assessment](https://skillstore.io/badges/skills/google-mantis-threat-model/security.svg)](https://skillstore.io/skills/google-mantis-threat-model?utm_source=security_passport_badge)

HTML 徽章

<a href="https://skillstore.io/skills/google-mantis-threat-model?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/google-mantis-threat-model/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

嵌入卡片

<iframe src="https://skillstore.io/embed/skills/google-mantis-threat-model.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

學術引用 (APA · BibTeX · CFF)

APA 引用

google. (2026). mantis-threat-model security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/google-mantis-threat-model/audits/1

BibTeX 引用

@techreport{google-google-mantis-threat-model-2026,
  author = {google},
  title = {mantis-threat-model security audit report (audit version 1)},
  institution = {Skillstore},
  year = {2026},
  number = {1},
  url = {https://skillstore.io/skills/google-mantis-threat-model/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: "mantis-threat-model security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
  - name: "google"
date-released: "2026-10-04"
url: "https://skillstore.io/skills/google-mantis-threat-model/audits/1"
identifiers:
  - type: other
    value: "skillstore:google-mantis-threat-model:audit:1"
    description: "Skillstore immutable audit report identifier"

Skillstore 評分

評分依據證據置信度: 中

架構

可維護性

內容

社群

規範符合度

你可以打造什麼

Review Architectural Exposure

Identify attacker entry points and trust boundaries from an existing Mantis architecture knowledge base.

Refresh Models After Snapshot Changes

Regenerate threat-model sections and reassess deployment intent when the recorded snapshot changes.

Prepare Security Planning Context

Provide downstream planning agents with attacker profiles, entity references, and asset availability classifications.

試試這些提示

Create an Initial Model

Run /mantis-threat-model using the populated Mantis knowledge base in the current workspace.

Use a Separate State Directory

Run /mantis-threat-model --state_root [state_directory]. Use its architecture and entity documents, archive the prior model, and save the replacement.

Refresh for a New Snapshot

Run /mantis-threat-model --state_root [state_directory] --snapshot_id [snapshot_id]. Recompute deployment intent when provenance changes and default to production for uncertain evidence.

Review Reuse and Archive Integrity

Run /mantis-threat-model --state_root [state_directory] --snapshot_id [snapshot_id]. Preserve existing archives; add the required STALE banner if unchanged provenance leads to reused content.

最佳實務

Populate architecture and entity documents with current deployment, entrypoint, and availability evidence before running the skill.
Keep snapshot identifiers consistent with the knowledge base revision and review models marked stale.
Preserve existing pass archives and verify the backup before replacing the live threat model.

避免

Using this skill as a source-code scanner or JSONL learning extractor.
Declaring a system sample-only when production evidence is missing, contradictory, or uncertain.
Treating unchanged snapshot labels as proof that the knowledge base is current or rerunning passes without archive protection.

常見問題

What inputs does this skill require?

It requires populated Mantis architecture and entity documents. Workflow state and an existing model provide optional pass and provenance metadata.

Does it inspect application source code?

No. This KB-only stage reads synthesized architecture and entity documents, not target source code or raw JSONL learnings.

Where is the threat model saved?

It saves THREAT_MODEL.md in the selected state workspace KB and copies the previous model into the pass archive.

What happens when snapshot metadata is missing?

Without an explicit snapshot identifier or pinned state metadata, it uses UNPINNED provenance and rederives the model.

How does it classify deployment intent?

It allows sample-only classification only when all five production-signal checks pass. Any uncertainty requires production classification.

Can repeated runs affect model history?

Yes. Reusing a pass number can replace its archived model. Preserve existing archives before rerunning the same pass.

開發者詳細資訊

作者

google

授權

MIT

Skillstore 修訂

版本提示

作者未聲明版本。

儲存庫

https://github.com/google/mantis/tree/f48a85e8823ee6f4824ae9a24b9b2efb8aab8c9a/mantis-threat-model

引用

afe56992452feed3201cea1348a69613bf5d0a6d

維護新鮮度

2026/10/5

使用量

0 次下載 · 0 次瀏覽

檔案結構

📄 SKILL.md