yotta-resume
"元简 — 本地求职者简历诊断、分析、修改与成品导出技能:分析简历结构、经历证据、量化指标、岗位匹配和风险,向用户追问缺失事实,给出逐条合理化修改建议,并把确认后的内容导出为 DOCX/PDF。触发:简历诊断、简历分析、改简历、简历追问、JD匹配、导出Word/PDF。边界:不编造经历,不承诺ATS、面试或Offer结果;不自动投递;不处理复杂设计作品集排版;默认本地处理且不联网。"
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/yottameta--yotta-resume)HTML
<a href="/listings/yottameta--yotta-resume"><img src="/api/listings/yottameta--yotta-resume/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Choose one of the methods below. Skill files are distributed through **npm**; GitHub hosts source code and issues.
Method 1: npm one-liner (recommended)
npx -y @yottameta/yotta-resume --agent <agent-name> npx -y @yottameta/yotta-resume --dir <agent-skills-dir>
- `--agent <name>` installs to that agent's default user-level skills directory. - `--dir <path>` installs to a specified skills directory. - `--list` shows supported agent defaults. - If a China mirror has not synced the package, append `--registry=https://registry.npmjs.org/`.
Method 2: git clone
git clone https://github.com/YottaMeta/yotta-resume.git <agent-skills-dir>/yotta-resume
Method 3: Download ZIP
Download the ZIP from `YottaMeta/yotta-resume`, then move the `yotta-resume` folder into your agent skills directory.
Method 4: install.sh
bash install.sh --agent <agent-name> bash install.sh --dir <skills-dir> bash install.sh --list
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
SKILL.md · Manifest · 116 lines
- ---
- name: yotta-resume
- description: "元简 — 本地求职者简历诊断、分析、修改与成品导出技能:分析简历结构、经历证据、量化指标、岗位匹配和风险,向用户追问缺失事实,给出逐条合理化修改建议,并把确认后的内容导出为 DOCX/PDF。触发:简历诊断、简历分析、改简历、简历追问、JD匹配、导出Word/PDF。边界:不编造经历,不承诺ATS、面试或Offer结果;不自动投递;不处理复杂设计作品集排版;默认本地处理且不联网。"
- version: 0.1.3
- license: MIT
- metadata:
- zh_name: "元简"
- ---
- # 元简(yotta-resume)
- 元简先根据简历内容自动识别行业,再分析结构、经历证据和 JD 关键词覆盖,
- 提出可回答的高价值问题,并按用户确认的事实导出 DOCX、HTML 或 PDF。
- ## 触发与边界
- - Triggers:用户要求诊断简历、改简历、补强经历证据、做 JD 匹配,或把已确认的简历导出为 Word/PDF。
- - Do NOT trigger:用户要求代投简历、编造经历或数字、预测面试结果、评估受保护个人属性,或制作复杂设计作品集排版。
- 常见自然语言说法:
- - “帮我看下这份安全运营简历哪里写虚了。”
- - “照这个 JD 看我还缺哪些关键词和证据。”
- - “先别改,告诉我还需要问我什么。”
- - “把确认后的内容导成 Word 和 PDF。”
- - “这段经历我只能确认范围,不能确认具体数字,应该怎么写?”
- 相近但不适用:“帮我海投”“把结果写成 90 分”“编一段大厂经历”“替我预测 Offer”。
- ## 核心流程
- 1. 首次安装或排查环境时先运行:
- ```bash
- python scripts/yotta_resume.py doctor
- ```
- `doctor` 会检查随包文件、数据契约、PyMuPDF、PDF 浏览器、临时目录和 Python 版本。
- PDF 依赖是可选项:缺少时仍可使用 Markdown / DOCX 输入和 DOCX / HTML 输出。
- 2. 读取简历文件。支持 UTF-8 Markdown、UTF-8 文本、DOCX 和文字层 PDF。
- PDF 解析优先使用本地 PyMuPDF;缺少依赖、加密、损坏或扫描件时明确失败或降级,
- 不生成伪完整报告。
- 3. 运行分析,优先让用户提供目标岗位 JD:
- ```bash
- python scripts/yotta_resume.py audit --resume resume.md --jd jd.txt --format json
- ```
- 4. 阅读 `source_manifest`、`parse_warnings`、`issues`、`unknowns` 和 `questions`。
- 先确认输入解析质量、规模、个人边界、结果、日期和 JD 缺口,再修改简历。
- 用户没有确认的事实不要写进简历。
- 5. 按用户确认的信息修改 Markdown,再导出成品:
- ```bash
- python scripts/yotta_resume.py export --input revised.md --docx revised.docx --html revised.html --pdf revised.pdf
- ```
- 6. 核对输出文件存在且可打开。多格式导出会先写入临时区,全部成功后一次性提交;
- 任一格式失败时不会留下本次部分成品,也不会覆盖原有同名文件。
- PDF 依赖本机 Chromium、Chrome 或 Edge;未找到浏览器时命令会明确失败,不会生成伪 PDF。
- 7. 需要单独校验随包岗位族、证据账本、题库、来源和双角色权限契约时运行:
- ```bash
- python scripts/yotta_resume.py contracts --format json
- ```
- ## 输出契约
- - `audit`:JSON 或 Markdown,包含文件来源清单、解析提示、行业识别、章节识别、
- 要点统计、工具发现、未知项、追问、修改建议、JD 匹配、结构准备度和边界声明。
- - `export`:DOCX / HTML / PDF。三者来自同一份 Markdown 内容模型,联系方式保留在成品简历中。
- - `contracts`:离线校验 `data/` 下的 6 个实际契约与 6 个 JSON Schema,
- 失败返回退出码 `6`,不写文件、不联网。
- - `doctor`:返回 `PASS / WARN / FAIL`,核心文件或契约失败返回退出码 `6`;
- PyMuPDF 或浏览器缺失只记为 `WARN`,不阻断 DOCX / HTML 路径。
- - 分析报告:手机号和邮箱会被打码;不承诺 ATS、面试或 Offer 结果。
- ## 评分与数字护栏
- - `readiness.score` 的展示名称固定为“结构准备度(规则分)”,
- `score_type` 为 `structure_rule_score`,只反映章节、量化证据、文本清洁度和
- 可选 JD 关键词覆盖,不是招聘方评分。
- - 只能原样引用工具返回的数值;**不得改写为其他数值、主观数字区间或“实际分数”**。
- 需要对招聘方视角作判断时,只允许定性标签并列出简历证据。
- - 未提供 JD 时 `jd_match.coverage` 必须为 `null`,报告写明不能判断岗位匹配。
- - `jd_match` 使用版本化固定词表,只检查明确关键词,不做语义匹配或岗位排名;
- 必须同时展示 `lexicon_version` 和 `limitations`。
- - 报告必须区分“工具规则发现”“智能体复核发现”“待用户确认事实”。
- 智能体额外发现不得冒充工具输出,且必须引用简历原文或来源页。
- - 规则不能判断招聘竞争力、面试通过率或 Offer 结果;必须列入 `unknowns`。
- ## 行业扩展
- - `industry_detection` 根据简历明确关键词识别行业,只使用可见内容,不读取受保护属性。
- - 当前可区分网络安全、软件研发、数据与人工智能、业务运营和通用兜底。
- - 网络安全已有 6 个岗位族和 18 道种子题;其他行业暂用通用证据与追问维度。
- - 新增行业时扩展 `data/industry-registry.json` 及 Schema,不需要修改主分析流程。
- - 多个行业信号接近时 `status=ambiguous`,不得自行猜测行业。
- ## 渐进披露
- - 详细输入格式、CLI 退出码和浏览器要求见 [references/usage.md](references/usage.md)。
- - 常见问题见 [references/faq.md](references/faq.md)。
- - 自然语言说法见 [references/trigger-examples.md](references/trigger-examples.md)。
- - 端到端工作流见 [references/workflow-examples.md](references/workflow-examples.md)。
- - 行业、岗位族、题库、来源与 HR 权限契约见 [references/data-contracts.md](references/data-contracts.md)。
- ## 自检
- ```bash
- python scripts/yotta_resume.py doctor
- ```
- 源码仓库维护者运行完整自动化测试;测试文件不作为终端用户运行前提。
README.md · README · 122 lines
- <p align="center"><b>Language</b>: English · <a href="./README.zh-CN.md">中文</a></p>
- <p align="center"><img src="./assets/banner.png" alt="yotta-resume banner" /></p>
- <h1 align="center">yotta-resume · Yuan Jian</h1>
- <p align="center">Analyze a resume locally, ask for missing evidence, suggest factual improvements, and export DOCX/PDF.</p>
- <p align="center">
- <a href="LICENSE"><img alt="License: MIT" src="https://img.shields.io/badge/license-MIT-blue" /></a>
- <a href="https://www.npmjs.com/package/@yottameta/yotta-resume"><img alt="npm package" src="https://img.shields.io/npm/v/@yottameta/yotta-resume" /></a>
- <a href="https://github.com/YottaMeta/yotta-resume"><img alt="GitHub stars" src="https://img.shields.io/github/stars/YottaMeta/yotta-resume" /></a>
- </p>
- ## What it is
- yotta-resume is a job-seeker resume analysis and export tool. It reads Markdown,
- UTF-8 text, DOCX, and text-layer PDF resumes, detects an industry from visible resume
- content, identifies sections, experience bullets, metrics, contact information, and
- JD keyword gaps, then returns high-value questions and improvement suggestions.
- After the job seeker confirms the facts, the same resume can be exported to DOCX,
- HTML, or PDF.
- The core analyzer uses the Python standard library. It makes no network requests,
- does not submit resumes, and never invents experience or numbers. PDF input uses
- local PyMuPDF and fails clearly when the dependency is missing or the file cannot
- be parsed.
- ## When to use
- - Review resume structure, evidence, and measurable outcomes.
- - Compare a resume with a target job description.
- - Prepare focused follow-up questions for missing facts.
- - Export a confirmed resume as DOCX/PDF.
- **Do not use it** to submit applications, fabricate experience, predict interviews or offers, evaluate protected personal attributes, or create complex design portfolios.
- ## Quick usage
- ```bash
- # Check the installed package and optional PDF capabilities
- python scripts/yotta_resume.py doctor
- # Analyze a resume and print a Markdown report
- python scripts/yotta_resume.py audit --resume resume.md
- # Analyze a text-layer PDF directly
- python scripts/yotta_resume.py audit --resume resume.pdf
- # Analyze with a JD and output JSON
- python scripts/yotta_resume.py audit --resume resume.md --jd jd.txt --format json
- # Export DOCX, HTML, and PDF
- python scripts/yotta_resume.py export --input revised.md --docx revised.docx --html revised.html --pdf revised.pdf
- # Validate role, question-bank, source, ledger, and dual-role contracts offline
- python scripts/yotta_resume.py contracts --format json
- ```
- PDF export requires a local Chromium, Chrome, or Edge browser. PDF input requires
- `pip install pymupdf`. Missing browsers, missing PDF dependencies, encrypted files,
- and image-only PDFs fail clearly instead of producing a misleading result.
- Multi-format export renders into a temporary area and commits only after every
- requested format succeeds, so a failed PDF does not leave new DOCX/HTML partial files.
- The `readiness.score` field is a structural rule score only. It is not a recruiter
- score. When no JD is provided, job-match coverage is `null`. JD matching uses a
- versioned fixed vocabulary and reports its limitations; it is not semantic matching
- or a hiring recommendation.
- Industry detection uses a versioned registry. Cybersecurity currently has the
- detailed role map and seed question bank; future industries can be added by extending
- `data/industry-registry.json` and its schema without changing the main analysis flow.
- See [references/usage.md](references/usage.md) for input format, fields, and exit codes.
- See [references/faq.md](references/faq.md) for common questions.
- See [references/data-contracts.md](references/data-contracts.md) for the bundled data contracts.
- ## Installation
- Choose one of the methods below. Skill files are distributed through **npm**; GitHub hosts source code and issues.
- ### Method 1: npm one-liner (recommended)
- ```text
- npx -y @yottameta/yotta-resume --agent <agent-name>
- npx -y @yottameta/yotta-resume --dir <agent-skills-dir>
- ```
- - `--agent <name>` installs to that agent's default user-level skills directory.
- - `--dir <path>` installs to a specified skills directory.
- - `--list` shows supported agent defaults.
- - If a China mirror has not synced the package, append `--registry=https://registry.npmjs.org/`.
- ### Method 2: git clone
- ```text
- git clone https://github.com/YottaMeta/yotta-resume.git <agent-skills-dir>/yotta-resume
- ```
- ### Method 3: Download ZIP
- Download the ZIP from `YottaMeta/yotta-resume`, then move the `yotta-resume` folder into your agent skills directory.
- ### Method 4: install.sh
- ```text
- bash install.sh --agent <agent-name>
- bash install.sh --dir <skills-dir>
- bash install.sh --list
- ```
- ## Validation
- Run `python scripts/yotta_resume.py doctor` after installation. It checks the bundled
- contracts, required files, optional PDF backends, browser availability, temporary
- directory access, and Python version. Full development tests live in the source
- repository and are not required to use the installed skill.
- ## License
- MIT © YottaMeta — see [LICENSE](./LICENSE).
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.