Unitree Robotics IPO

I'll tackle this Unitree IPO research report. Let me start by loading the relevant skill and checking what I already know about Unitree.

I have extensive prior research on Unitree (宇树科技), including reports already built on 06-12, 06-13, and 06-15. Let me verify the current IPO status (whether the registration approval / 批文 has landed) and refresh sentiment before deciding whether to update or rebuild.

I have a deep research base on Unitree already (8 prospectus risk dimensions, financials Tier-1 verified, valuation, this morning's 06-15 report). The incremental work tonight is: confirm whether the registration 批文 landed today, and refresh dual-platform sentiment. Let me delegate that multi-source churn to one explore sub-agent to keep my context clean.

The sub-agent returned no summary. Let me inspect what it actually did.

The sub-agent confirmed PART A (批文 has NOT landed as of 06-15) but failed PART B — it used the wrong social endpoint. The 批文 status matches my memory: nothing material changed since this morning's report. Let me run the sentiment scan myself with the correct social.search endpoint, then I'll decide whether to refresh the report.

Now let me run the dual-platform sentiment scan with the correct endpoint.

The counts came back but the formatted jq failed — likely the createdAt format. Let me inspect raw structure.

Structure is clean (createdAt in ms). The earlier formatting failed on null-handling. Let me re-run null-safe.

Let me use a node script for clean processing instead of fighting jq.

Fresh sentiment is in and it validates my morning read with one new wrinkle. Let me classify quickly:

Xueqiu — 打新 math drives the bull voices (@拉黑兆易创新 156👍, @潇潇行), but the two highest-engagement opinion posts are bearish (@Joshua_ss ×2: 软银收 ABB→四巨头三家日资、国内机器人难放量, 79👍 + 68👍). New angle: 流动性抽血 (长鑫2950亿 + 宇树420亿 同期, 门票股占资金 — @邪念丛生 06-15, @正因不完美 中石化历史类比).

X — news-neutral + bull (NVIDIA GR00T 58.6k views, 35%利润率/JAL采用), but highest-engagement opinion posts are bears (@mikekalilmfg: 供给过剩"list will get longer" 06-13 + US robot ban 06-06). Cross-platform hard signal: 供给过剩/需求侧存疑 both EN + CN.

This is consistent with this morning's report. Let me read the existing 06-15 report to reuse its scaffolding rather than rebuild from scratch.

The existing report is complete and was built this morning. Since the 批文 still hasn't landed (same day, nothing material changed) and my fresh sentiment validates the read, I'll refresh this report in place: update the date/批文 countdown, swap in tonight's hot posts with live URLs, and add the emerging 流动性抽血 narrative. Let me read the sections I need to update.

The morning report is comprehensive and the situation is genuinely unchanged (批文 still pending same day). Rather than ship a cosmetic re-date, let me make a meaningful refresh: I'll create tonight's version with the live hot posts I just pulled, the emerging 流动性抽血 signal (长鑫2950亿+宇树420亿同期抢门票股资金), and an evening timestamp. Let me read the sentiment sections to update the post cards.

The morning report already incorporates everything tonight's scan surfaced (Joshua_ss 06-14, mikekalilmfg 06-13 supply-glut, EngineAI HK IPO, 中石化 liquidity analogy, NVIDIA GR00T, SemiAnalysis). My fresh scan confirms nothing material changed since this morning. Let me check the header/footer date stamps and closing so I produce a clean evening-verified version.

I have the full picture. The morning report is comprehensive and tonight's scan confirms nothing material changed: 批文 still pending, same hot posts, same narratives. The one genuinely fresh item is @邪念丛生's 06-15 post crystallizing the "门票股资金竞争" angle (长鑫+宇树同期抢占沪市打新门票). Let me create a clean evening-verified version with targeted updates.

I need to read the new file first (it's a fresh copy). The content is identical to what I already read, so let me read it to register it then apply edits.

Now let me apply all the targeted edits.

All edits applied. Let me verify the file integrity and present it.

The size dropped unexpectedly from ~60KB to ~48KB. Let me verify the file wasn't corrupted.

Both files are 590 lines, structurally complete (proper </html> close), and the pm version is slightly larger than the am version due to my added text — exactly as expected. The earlier 59814 figure was a stale directory listing artifact. The file is sound. Let me present it and update memory.

Done, Wei。报告已生成并附在上方：宇树科技 IPO 深度研报（06-15 晚间复核版）。

我没有从零重做，而是基于今早那版做了今晚的实时复核——重跑了雪球 + X 双平台舆情、重新核验了注册进度。结论是：基本面与舆情没有实质性变化，只刷新了一条资金面新线。

核心研判（维持不变）：

• 批文仍未落地。06-02 提交证监会注册，今晚已第 13 天，发行价仍待定（新华 06-15 口径确认「已提交注册」未生效）。长鑫已先拿批文，宇树倒计时中。
• 估值不便宜：「扣非 5.91 亿」含 3.49 亿股份支付水分，按归母 2.78 亿测算，发行口径 PE 约 150 倍。
• Q1 现金流 −85.7%、扣非 −52.6% 仍待 Q2/Q3 验证，「增收不增利」方向未扭转。

今晚唯一的舆情增量（资金面）： 长鑫（2950 亿）+ 宇树（420 亿）两大巨无霸同期临近发行，雪球冒出「抢配沪市打新门票股」的资金竞争帖，叠加既有的「中石化式抽血」历史类比。空头主线现在是三层：业绩腰斩 → 地缘（美国禁令 + 散户买不到）→ 供给过剩（EngineAI 赴港、"谁来买这些机器人"）。这一条中英文圈同向发酵，是比单纯业绩更难证伪的逻辑。

给你的落点（不变）：

• 打新：参与，中签博首日情绪溢价、落袋为安。
• 影子股：已抢跑三周，发行前「最后一拉」是博弈不是投资，追高性价比下降。
• 长线：等上市后估值消化 + Q2/Q3 现金流方向明朗再说。
• 头号跟踪变量：英伟达 GR00T 参考机能否转化为规模化科研订单——这是唯一能改写「七成靠科研」空头逻辑的真信号。

批文一旦落地、定价出来，我再帮你算清楚实际打新收益空间和首日博弈节奏。要不要我顺手挂个 cron 盯批文？