# Interview Minutes Analysis Payload Builder

> Runs the interview-minutes analysis flow end to end — scans the ATS Bitable, reads the Lark Minutes transcript, analyzes the interview with your own model, writes the AI notes back to the ATS record, and posts the HR card.

- Skill: `chronoaiproject/interview-minutes-analysis-payload-builder` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx get-skill add chronoaiproject/interview-minutes-analysis-payload-builder`
- Raw SKILL.md: https://api.skillmd.com/api/skills/chronoaiproject/interview-minutes-analysis-payload-builder/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: ChronoAIProject (https://skillmd.com/u/chronoaiproject)
- Updated: 2026-11-16
- Page: https://skillmd.com/skills/chronoaiproject/interview-minutes-analysis-payload-builder

---


# Interview Minutes Analysis

Use this when someone asks to analyze pending Lark Minutes interview records and notify HR.

**You (the agent) run the WHOLE flow yourself with your own tools, then WRITE the results to Lark
Bitable and post a summary to the chat.** Do NOT refuse, do NOT just describe, do NOT claim "the
skill only formats" — drive the flow. If a data source or the target base is unavailable, SKIP that
part and still post what you have. Never ask the user for tokens — your NyxID-brokered tools broker
all credentials.

## How to run it

1. **Resolve inputs and target.** Use the flow's ATS Bitable and HR recipient unless the user
   overrides them: target Bitable `app_token` `FSl0bCi9raBuLbsdTbHlgb0agwf`, `table_id`
   `tblgZgSqmeBag2na`; HR `open_id` `ou_4473a6a9feeca68d2eeaf2cb95416f37`. If the user named a
   specific record or Minutes URL, use it; otherwise process the first pending ATS record. If the
   ATS base is unresolvable, proceed in summary-only mode.

2. **COLLECT the ATS record.** Read the ATS Bitable via `nyxid_proxy` `{slug:"api-lark-bot",
   path:"/open-apis/bitable/v1/apps/<app_token>/tables/<table_id>/records?page_size=100&filter=<urlencode('CurrentValue.[Minutes URL]!=\"\"')>",
   method:"GET"}` (or the equivalent `lark_*` tool). Pick the first item whose `Minutes URL`
   contains `/minutes/` and whose `AI Interview Notes` is empty. Keep `record_id`, `Minutes URL`,
   `Candidate Name`, `Job Title`, `Job Description`, and `Interviewer`. If there is no pending
   record, post that and stop.

3. **COLLECT the Minutes transcript.** Extract `minute_token` with `/minutes/([a-z0-9]+)/i`, then
   read Minutes via `nyxid_proxy` slug `api-lark-bot` (or `lark_minutes`):
   `/open-apis/minutes/v1/minutes/{minute_token}` and
   `/open-apis/minutes/v1/minutes/{minute_token}/transcript?need_speaker=true&need_timestamp=true&file_format=txt`.
   Use the returned title, else `<candidate_name> 面试`. If the transcript is empty, post that it may
   still be processing and stop.

4. **ANALYZE.** With your own model, classify the role as `技术岗`, `商业/职能岗`, or `家政/个人助理岗`
   from the title/JD/transcript, run the role-specific veto checks, and return valid JSON with
   `veto_check`, `overall`, `score`, `summary`, `strengths`, `red_flags`, `key_moments`,
   `culture_fit`, `culture_fit_reason`, `next_step`, `suggested_questions`, and `veto_count`.
   - 商业/职能岗: AI 实质改造、具体 AI 案例、独立贡献、从 0 到 1、思维深度。
   - 技术岗: AI 编程实质、Agent/AI 技术栈、系统设计能力、独立交付能力，`v5` 填 `N/A`。
   - 家政/个人助理岗: 诚信核查、稳定性、核心技能，`v4` 和 `v5` 填 `N/A`。
   Any VETO caps `score` at 2; multiple VETOs normally mean `score: 1` and `overall: "Strong Pass"`.
   Base every judgment on transcript evidence.

5. **BUILD the exact payloads.** Run `scripts/build_interview_analysis_payload.js` via `code_execute`
   (language `"javascript"`; NO stdin — inline the data into the code):
   `console.log(JSON.stringify(buildPayload({ mode: "build_payload", analysis, hrOpenId, atsAppToken,
   atsTableId, recordId, generatedAt, candidateName, jobTitle, minuteTitle })))`, where `generatedAt`
   is the current Asia/Singapore timestamp. Use the output's `lark.bitableUpdate.body` and
   `lark.cardMessage.body`. If code execution is unavailable, follow
   `references/interview-analysis-contract.md` field-for-field.

6. **WRITE the notes, then post the card.** Update the ATS record via `nyxid_proxy`
   `{slug:"api-lark-bot",
   path:"/open-apis/bitable/v1/apps/<app_token>/tables/<table_id>/records/<record_id>", method:"PUT",
   body:<lark.bitableUpdate.body>}`. THEN send the HR interactive card with `lark_messages_send` (or
   `reply_with_interaction`); the equivalent is `nyxid_proxy`
   `POST /open-apis/im/v1/messages?receive_id_type=open_id` with `lark.cardMessage.body`. **If no ATS
   target was resolved OR the update fails, SKIP the Bitable write and STILL post the analysis card /
   summary to the chat — never fail the run for a missing base.**

7. **REPORT** one line: candidate, job title, score/overall, ATS record updated (or "summary-only"),
   HR card sent.

## Payload builder reference

Bundled script: `scripts/build_interview_analysis_payload.js`. Accepts `mode: "build_payload"` (or
omitted); other modes return `skip: true`. Required after alias normalization: `analysis`,
`hrOpenId`, `atsAppToken`, `atsTableId`, `recordId`, `generatedAt`. Optional: `candidateName`,
`jobTitle`, `minuteTitle`. `analysis` may be a direct object or a Groq-style
`choices[0].message.content` string. Output includes `notes_text`, `card`, `lark_card_json`,
`lark.bitableUpdate` (PUT), and `lark.cardMessage` (POST to `open_id`). Exact aliases, fallback
parsing, card fields, and request bodies are in `references/interview-analysis-contract.md`.

## Guardrails

- Only use the Lark records, Minutes metadata, transcript text, and analysis evidence the tools
  actually returned — never invent ids, records, statuses, quotes, or candidate details.
- Prefer running the bundled script; it is the deterministic source of truth for payload shape.
- All Lark access, sends, and credential exchange go through your NyxID-brokered tools or Lark
  tools; never handle raw secrets.

