interview-processor

Agent Skill: interview-processor (Interview Record Processing Skill)

This skill is used to assist interviewers in efficient structured processing before and after interviews, consolidate recruitment results, and objectively self-assess questioning proficiency. It can work with memories-off skill's memocli to safely write generated data into the local knowledge graph.

1. Dependencies and Specification Declaration (Dependencies)

This skill depends on the local-first knowledge base management tool memocli。

External Dependencies and Specification Description

This Skill depends on the memories-off library for entity management and long-term memory. Before executing any task, you must first consult and fully follow the memories-off-declare.md declaration document in the current directory to obtain the entity type specifications and encapsulated sub-process operation details defined therein.

At the same time, to simplify the input length of instructions, it is recommended that you configure a global path alias in ~/.config/memocli/config.yaml and use abbreviated aliases (such as -p work or -p life) to run all memocli instructions involved in this skill.

---

2. Dynamic Context Loading Mechanism (Interviewer Persona)

To ensure that the "team introduction" in interview summaries and the self-assessment of the interviewer's performance align with the interviewer's current latest position and team, this skill adopts aknowledge graph prioritydynamic persona loading mechanism:

1. Step 1: Before executing any subtask, the Agent must run:

   memocli search-entities "我自己的面试官画像" -p <path>
   

2. Step 2：
   • If the entity exists, the Agent should read its body (usually containing ## 当前岗位与团队 and ## 个人面试偏好), and use its content as the current background context.
   • If the entity does not exist, the Agent should kindly remind the interviewer and guide them to initialize the persona using the following command:

     echo "## 当前岗位与团队
     [在此写明您当前的岗位、所负责团队及核心业务介绍]
     
     ## 个人面试偏好
     [在此写明您的面试风格、提问原则和关注特质]" | memocli create-entity -p <path> -e "我自己的面试官画像" -t "个人画像" --content-stdin --reason "初始化面试官个人画像"
     

---

3. Data Entity Modeling Scheme (KG Entity Models)

This skill manages four types of entities in the local knowledge base. To comply with the memories-off specification, all entities must follow strict H1 (# 实体名) and H2 (## 章节名) heading levels, and H3 and lower-level headings are prohibited.

3.1 Interviewer Persona

• Naming Convention：我自己的面试官画像
• Entity Type (type)：个人画像
• Core Sections：
  • ## 当前岗位与团队
  • ## 个人面试偏好

3.2 Interview Methodology

• Naming Convention：[领域/岗位]专家面试风格-YYYYMMDD (for example:iOS专家面试风格-20250718）
• Entity Type (type)：面试方法论
• Core Sections：
  • ## 面试风格与偏好(list the interviewer's questioning principles, style, and follow-up logic in this field)

3.3 Interview Question Set

• Naming Convention：[领域/岗位]面试问题集-YYYYMMDD (for example:大模型产品面试问题集-20250729）
• Entity Type (type)：面试问题集
• Core Sections: multiple H2 sections divided according to the interview direction or technical module.
  • ## 核心认知与边界理解
  • ## 系统设计与成本控制
  • ... (dynamically created according to the actual direction)
• Outbound Relationships：
  • target_methodology points to its corresponding [领域/岗位]专家面试风格-YYYYMMDD entity.

3.4 Interview Record

• Naming Convention：对[候选人姓名]的面试总结 (YYYY-MM-DD)
• Entity Type (type)：面试记录
• Core Sections：
  • ## 基本信息(candidate name, recruitment type [campus recruitment/social recruitment/Digital Mali], position name, interview date)
  • ## 面试反馈(recommended level, strengths, weaknesses, conclusion)
  • ## 沟通记录与评价(Core discussion topic list and candidate response assessment)
  • ## 面试官表现自评(Objective improvement requirements for interviewers' questions, follow-ups, and appropriateness at an expert level)
• Outbound relationships：
  • use_style points to the corresponding 面试方法论 entity.
  • use_question_set points to the corresponding 面试问题集 entity.
  • candidate points to the candidate's personal entity (if any).

---

4. Subtasks Navigator

When the user triggers this skill, please jump to the corresponding subtask execution detailed guide based on the current interview stage:

• Before the interview：Subtask 1: Plan the interview outline — Generate a customized interview outline based on the resume and JD.
• After the interview (question organization)：Subtask 3: Organize real questions — Extract, correct, and standardize all the interviewer's real questions from the recording transcript.
• After the interview (minutes summary)：Subtask 2: Organize interview minutes and evaluation — Generate an interview summary report, conduct an expert-level self-assessment of the interviewer's performance, and write it into the knowledge graph.