SuperAPI AI Directory
Partner中文AI newsFavorites
Skillsanalyze-resume

analyze-resume

Coding & development

View repository

Updated 10-16 09:57

About this skill

Analyze Resume — Comprehensive Resume Analysis

Input a resume file (PDF or text), automatically extract candidate information, and output a structured interview analysis report.

When to Use

Triggered when the user requests the following operations:

  • "Analyze this resume" / "Help me take a look at this resume"
  • "analyze resume" / "review this CV"
  • "How is this candidate" / "Help me come up with interview questions"
  • "Resume analysis" / "Candidate evaluation"

Applicable positions: Software Engineer, Test Development Engineer, Frontend Development Engineer

Phase 0: Read the resume

  1. The user provides a file path (PDF or text file)
  2. Use the Read tool to read the file contents
  3. If it is a PDF, read it directly with the Read tool (PDF parsing supported)
  4. After confirming the read is successful, proceed to the analysis phase

Phase 1: Project experience summary

For each project/work experience of the candidate, extract the following dimensions:

Dimension Description
What was done The candidate's personal contribution, focusing on "I" rather than "the team"
Role Core developer / Lead / Participant / Support role
Difficulty ⭐~⭐⭐⭐⭐⭐, assessed based on technical complexity, scale, and innovation
Highlights The most noteworthy technical highlights or achievements
Shortcomings Vague descriptions, unclear personal contributions, lack of quantitative data, etc.

Output format:

### [项目名称] @ [公司/学校] ([时间])
- 做了什么：...
- 角色：...
- 难度：⭐⭐⭐
- 亮点：...
- 不足：...

Phase 2: Comprehensive assessment (not limited to technical/time/expression and collaboration clues)

Conduct a comprehensive assessment from the three dimensions of technical ability, reasonableness of time arrangement, and expression and collaboration clues, and output the following three points:

🔴 Highly questionable resume descriptions (1 point)

Select the most suspicious and most worth questioning description, which may be:

  • Exaggerated technical description ("led" but appears to be only participation)
  • Unreasonable achievement data (99% performance improvement but no explanation)
  • Contradictory timeline or abnormal density of project experience
  • Mismatch between tech stack and responsibilities

Format:[描述原文] → 质疑点：...

⚠️ Most noteworthy shortcoming (1 point)

The candidate's most prominent weakness, which may be:

  • Insufficient technical depth (only knows how to use, not the principles)
  • Single project experience (industry/tech stack too narrow)
  • Lack of system design or architecture experience
  • Unclear expression, questionable communication ability

✅ Most noteworthy strength (1 point)

The candidate's most prominent highlight, which may be:

  • Technical breadth or depth clearly higher than peers
  • Has actually launched projects or open-source contributions
  • Has solved real high-difficulty problems
  • Strong learning ability and fast technical iteration

Phase 3: Technical expertise (3 points)

Not limited to a single project; synthesize all experience to extract the candidate's core technical expertise:

1. [专长领域]：[具体说明，引用简历依据]
2. [专长领域]：[具体说明，引用简历依据]
3. [专长领域]：[具体说明，引用简历依据]

Selection criteria:

  • Technologies that appear repeatedly across multiple projects
  • Technical directions described most specifically and with the most depth
  • Core abilities most relevant to the target position

Phase 4: Assessment Question Generation

Based on the 3 specialties from Phase 3, selectthe 2 most discriminative directions, and generate one set of progressively in-depth assessment questions for each:

Assessment Direction Format

## 考察方向一：[专长方向]

**Q1（基础）**：[问题]
参考答案：[简要答案，100字内]

**Q2（进阶）**：[问题]
参考答案：[简要答案，150字内，可含代码片段]

**Q3（深入）**：[问题]
参考答案：[简要答案，200字内，考察原理/边界/取舍]

Progression Logic:

  • Q1: Conceptual understanding, confirm the candidate has mastered the basics
  • Q2: Practical application, assess the ability to solve real problems
  • Q3: Principles/boundaries/best practices, distinguish excellent candidates from ordinary ones

Phase 5: LeetCode Algorithm Question Recommendations

Based on the candidate's project experience and technical specialties, find algorithm questions that havea certain connectionto their background (rather than generic popular questions), one each at low/medium/high difficulty:

Relevance Principles:

  • Candidate has worked on distributed systems → recommend question types involving concurrency and consistency
  • Candidate has worked on search/recommendation → recommend graph, sorting, and heap-related question types
  • Candidate has worked on data processing/ETL → recommend string, sliding window, and hash-related question types
  • Candidate has worked on Android/mobile → recommend design patterns, trees, and state machine-related question types

Output format:

### 🟢 简单：[题目编号]. [题目名称]
- 题目链接：https://leetcode.cn/problems/[slug]/
- 关联理由：[为何与候选人经历相关]
- 考察点：[核心算法/数据结构]
- 题目描述：
  > [原题完整描述，原样照录，不简化，包含所有约束条件和边界说明]

  **示例：**

Input: [original problem sample input, reproduced in full] Output: [original problem sample output, reproduced in full] Explanation: [original problem explanation, reproduced in full; must be included if the original problem has one]

- 解题思路：[关键步骤，2-3句]

[pseudocode, 5-10 lines, reflecting the core logic]

- 时间复杂度：O(?) / 空间复杂度：O(?)

### 🟡 中等：[题目编号]. [题目名称]
- 题目链接：https://leetcode.cn/problems/[slug]/
- 关联理由：[为何与候选人经历相关]
- 考察点：[核心算法/数据结构]
- 题目描述：
> [原题完整描述，原样照录，不简化，包含所有约束条件和边界说明]

**示例：**

Input: [original problem sample input, reproduced in full] Output: [original problem sample output, reproduced in full] Explanation: [original problem explanation, reproduced in full; must be included if the original problem has one]

- 解题思路：[关键步骤，2-3句]

[pseudocode, 5-10 lines, reflecting the core logic]

- 时间复杂度：O(?) / 空间复杂度：O(?)

### 🔴 困难：[题目编号]. [题目名称]
- 题目链接：https://leetcode.cn/problems/[slug]/
- 关联理由：[为何与候选人经历相关]
- 考察点：[核心算法/数据结构]
- 题目描述：
> [原题完整描述，原样照录，不简化，包含所有约束条件和边界说明]

**示例：**

Input: [original problem sample input, reproduced in full] Output: [original problem sample output, reproduced in full] Explanation: [original problem explanation, reproduced in full; must be included if the original problem has one]

- 解题思路：[关键步骤，2-3句]

[pseudocode, 5-10 lines, reflecting the core logic]

- 时间复杂度：O(?) / 空间复杂度：O(?)

Note: only recommend real problems that you are certain exist on LeetCode; the problem number and slug must be accurate; do not fabricate.

Phase 6: Programming Language-Specific Assessment

Based on the programming language the candidate uses most or is best at in their resumeuses most or is best at, create 3 syntax or professional questions from easy to hard, assessing true mastery of language features (rather than algorithm ability):

Question Type Directions (examples by language):

  • Go: goroutine/channel usage, defer execution order, interface implementation, slice internals, context propagation
  • Java: GC mechanisms, concurrency utility classes, generic type erasure, reflection, JVM memory model
  • Python: GIL, decorators, generators, mutable default argument pitfall, metaclasses
  • JavaScript/TypeScript: event loop, closures, prototype chain, Promise, type system
  • C++: RAII, move semantics, virtual function table, template specialization, memory management

Output format:

### 编程语言：[语言名称]

**Q1（基础）**：[问题]
解答思路：[2-3句，说明正确答案及关键原理]

**Q2（进阶）**：[问题]
解答思路：[3-4句，含代码片段或具体场景说明]

**Q3（深入）**：[问题]
解答思路：[3-5句，考察底层原理、边界情况或与其他语言的对比]

Question Design Principles:

  • Each question targets a specific language feature; avoid broad "talk about your understanding of XX"
  • Prioritize assessing language features the candidateactually usedin their resume (e.g., candidate mentions using goroutine for concurrency → ask channel close/select-related questions)
  • Q3 should have a certain degree of discrimination, able to distinguish candidates who have "used it" from those who "understand the underlying mechanisms"

Output Structure

# 简历分析报告：[候选人姓名]
> 岗位方向：[软件工程师 / 测试开发 / 前端开发]
> 分析时间：[日期]

---

## 一、项目经历小结
[Phase 1 输出]

---

## 二、综合评估
[Phase 2 输出]

---

## 三、技术专长
[Phase 3 输出]

---

## 四、面试考察题目
[Phase 4 输出]

---

## 五、LeetCode 算法题推荐
[Phase 5 输出]

---

## 六、编程语言专项考察
[Phase 6 输出]

Notes

  • Remain objective: Based on analysis of resume content, do not make unsupported inferences
  • Low-confidence clues must be labeled: Judgments involving expression, collaboration, and communication can only be weak conclusions based on how the resume is presented, and cannot be treated as definitive facts
  • Questioning must be evidence-based: The questioning points in Phase 2 must cite the original text of the resume
  • Questions must be actionable: Assessment questions must be specific; do not ask broad, vague major questions
  • Reference answers must be precise: Reference answers should have clear judgment criteria to facilitate interviewer scoring