技能 mantis-history
📦
mantis-history
内容修订 r1 中风险 ⚙️ 外部命令📁 文件系统访问
Extract Historical Security Fixes with Mantis
Past security fixes are difficult to track across repository history. This skill guides an AI agent to extract, classify, and cache historical vulnerabilities for later analysis.
支持: Claude Codex Code(CC)
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 "mantis-history" from https://skillstore.io/skills/google-mantis-history.md and its manifest at https://skillstore.io/api/skills/google-mantis-history/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.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“mantis-history”。 Illustrative request: Extract historical authentication fixes from this repository.
预期结果:
- Example finding: An authentication check was added before accessing an account resource.
- The historical record links the revision, affected file, vulnerability type, and mitigation.
- The conclusion requires verification against the actual revision diff.
正在使用“mantis-history”。 Illustrative request: Analyze security history from a shallow Git clone.
预期结果:
- History status: PARTIAL_SHALLOW.
- Available revisions were analyzed; older history remains unavailable.
- Missing historical findings do not establish that a component is secure.
正在使用“mantis-history”。 Illustrative request: Analyze a directory without accessible VCS history.
预期结果:
- History status: UNSUPPORTED_VCS.
- An empty historical learnings database was written.
- No historical vulnerabilities were fabricated.
安全审计
中风险v1 • 2026/10/4 打开版本化报告
All 58 static findings are false positives involving Markdown, legitimate workspace access, or read-only repository checks. A separate semantic risk concerns sending repository diffs and messages to unspecified LLM services without data-sharing safeguards. No evidence found of malicious commands, credential theft, or an actual prompt injection attempt.
1
已扫描文件
240
分析行数
0
复核项
0
已忽略误报
已确认安全问题 (1)
中
Repository Data Sharing Without Defined Safeguards
The skill requests LLM calls containing commit diffs and messages without specifying provider approval, secret redaction, or local-only processing. External implementations could disclose proprietary code or historical secrets.
The instructions explicitly batch diffs and messages into LLM calls but define no data-sharing controls. Actual disclosure depends on the selected execution backend.
风险因素
⚙️ 外部命令 (45)
SKILL.md:20 SKILL.md:23 SKILL.md:25 SKILL.md:27 SKILL.md:33 SKILL.md:36 SKILL.md:40 SKILL.md:42 SKILL.md:43 SKILL.md:44 SKILL.md:50 SKILL.md:51 SKILL.md:52 SKILL.md:58-97 SKILL.md:97-99 SKILL.md:99 SKILL.md:103-104 SKILL.md:104-105 SKILL.md:105-116 SKILL.md:116-127 SKILL.md:127 SKILL.md:128 SKILL.md:130-131 SKILL.md:131-132 SKILL.md:132-133 SKILL.md:133-135 SKILL.md:135-136 SKILL.md:136-140 SKILL.md:140-166 SKILL.md:166-167 SKILL.md:167-168 SKILL.md:168-169 SKILL.md:169-170 SKILL.md:170-173 SKILL.md:173-174 SKILL.md:174-176 SKILL.md:176-177 SKILL.md:177-178 SKILL.md:178-179 SKILL.md:179-206 SKILL.md:206-209 SKILL.md:209-211 SKILL.md:211-230 SKILL.md:230-234 SKILL.md:234-236
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/google-mantis-history/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/google-mantis-history?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/google-mantis-history?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/google-mantis-history/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/google-mantis-history.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
google. (2026). mantis-history security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/google-mantis-history/audits/1BibTeX 引用
@techreport{google-google-mantis-history-2026,
author = {google},
title = {mantis-history security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/google-mantis-history/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-history 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-history/audits/1"
identifiers:
- type: other
value: "skillstore:google-mantis-history:audit:1"
description: "Skillstore immutable audit report identifier"
Skillstore 评分
评分依据 证据置信度: 中 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
Prepare Historical Security Context
Build a repository-specific database of earlier vulnerabilities and mitigations before a broader security investigation.
Track Recurring Vulnerability Patterns
Review historical fixes by component and vulnerability type to identify recurring maintenance risks.
Feed Downstream Mantis Analysis
Generate state-relative historical learnings with cached records and provenance for subsequent analysis stages.
试试这些提示
Beginner: Summarize Security History
Use mantis-history to extract past security fixes from this repository. Report the output location and any history limitations.
Intermediate: Tailor History Filters
Inspect this repository's stack and choose relevant security keywords. Extract historical findings and explain which commits your filters may exclude.
Advanced: Refresh Cached Learnings
Refresh historical learnings from the existing cache. Verify revision reachability and repository identity; rebuild the cache if history was rewritten.
Expert: Coordinate Snapshot Analysis
Use mantis-history with --snapshot_root <snapshot>, --snapshot_id <id>, and --state_root <state>. Read live VCS history and preserve snapshot provenance.
最佳实践
- Approve the analysis backend and redact sensitive repository content before any external LLM processing.
- Keep scripts, caches, and outputs under the designated state workspace, outside the target tree.
- Verify extracted findings against their revisions and record partial or unsupported history explicitly.
避免
- Treating an empty or filtered history database as proof that the repository has no vulnerabilities.
- Reusing cached findings after history rewrites without checking revision reachability and repository identity.
- Sending private diffs or historical secrets to an unapproved external analysis service.
常见问题
Which AI tools are listed as supported?
The report lists Claude, Codex, and Claude Code. Extraction also requires repository access and an environment capable of running generated scripts.
Does this skill review or patch current code?
No. It extracts historical vulnerabilities and fixes. Its stated scope excludes code reviews, test-script creation, and patching.
Where are the generated files stored?
With --state_root, scripts, caches, and historical_learnings.jsonl belong under its workspace directory. Without locator arguments, the workflow uses the live working directory.
How does it handle missing or shallow history?
Missing or unsupported history produces an empty database marked UNSUPPORTED_VCS. Shallow Git history is analyzed with the PARTIAL_SHALLOW marker.
How are repeated runs handled?
The output is rebuilt from cached records. New revisions are analyzed incrementally, and rewritten history or changed repository identity invalidates the cache.
Can repository content reach an external service?
Yes, if the selected LLM backend is external. The skill requests diff and message analysis but does not define redaction or provider approval.