Disclaimer — please read first. This Agent provides informational and
educational analysis only. It does not constitute professional financial
advice. You should consult a licensed financial advisor before making any
financial decision. Reports contain no price targets, trading instructions,
or position-size recommendations. The publisher is not a licensed investment
advisor. Past performance and historical patterns do not guarantee future
results. Always do your own research.

RocketStock Deep Fundamental Analysis

Most stock skills summarize other people's opinions. This one runs a real
research pipeline on a dedicated backend and returns a full analyst-grade
report in English.

What you get

• Valuation you can audit — dual-track DCF (current capex vs normalized),
  cross-checked three ways against analyst consensus and composite verdicts,
  with sensitivity ranges instead of single-point guesses.
• Forensic accounting — an earnings-quality audit (does reported profit
  actually convert to cash?) and the 8-variable Beneish M-score manipulation
  screen. Almost no retail product does this.
• Management accountability — guidance history is stored and scored: does
  this management underpromise and overdeliver, or the opposite? Earnings-call
  analysis included.
• A watchlist with memory — every report ends with concrete, quantified
  watch items that are re-checked automatically against the next earnings
  print. Your analysis persists across quarters.
• Honesty engineering — the report inventories exactly which data was and
  wasn't available on that run, quarantines unverified news figures, and flags
  suspicious inputs instead of papering over them.
• Taiwan coverage nobody else has — for .TW / .TWO tickers the backend
  reads MOPS statutory disclosures, official monthly revenue (Taiwan's uniquely
  early leading indicator), director/supervisor share pledges, institutional
  flows and Chinese-language earnings calls — and writes it all up in English.

How to use it

Just name a ticker:

• US stocks — NVDA, PLTR, BRK.B
• Taiwan stocks — 2330.TW (TWSE) or 6488.TWO (TPEx)

The Agent submits the job, waits for the backend to finish, and returns the
report. A first run of the day typically takes 10–15 minutes, and can
occasionally reach 20 when other analyses are queued ahead of yours; re-opening
a ticker you already ran the same day is instant. Afterwards you can ask
follow-up questions grounded in the report.

Coverage is US and Taiwan listings only. Other markets (Hong Kong, London,
Tokyo, mainland China, Korea) are not supported; where a US-listed ADR exists,
ask for that instead.

Output format

Two parts. In the chat you get a concise summary — a one-line verdict, the
valuation range (local DCF vs. sell-side consensus vs. current price), anything
that changed since the previous run, and the disclaimer. The full report
arrives as two file attachments carrying the same content: a Markdown
file you can keep, diff against next quarter's, or open in any editor, and a
self-contained HTML page that opens in any browser with the full formatting
— no internet connection required, nothing loaded from anywhere else. Both
contain sections 0–11 of the deep analysis, the complete analytical conclusion,
the next-earnings watchlist table, data-consistency checks and a data-coverage
inventory.

A first analysis of the day takes roughly 10–15 minutes; the Agent waits and
delivers everything in the same reply. A ticker already analysed that day comes
back in seconds.

Quota

Two separate limits, and they count different things:

• 40 conversation turns per week — the platform-level cap shown on the plan.
  Every message you send counts, including follow-up questions about a report you
  already have. That is comfortably above what a typical week needs: two full
  analyses a day plus several follow-ups on each.
• Up to 2 full analyses per day, enforced server-side. This is the cap on
  fresh research runs, and it is the one that governs new coverage.

Follow-up questions, and re-reading a report that already exists for the current
day, cost a conversation turn but do not consume the analysis quota. If the
daily analysis quota is exhausted, the Agent tells you plainly and you can
continue the next day.

Data & privacy

• Two things leave the container: the ticker symbol you ask about, and a
  random quota identifier the client generates on first run. The identifier
  contains no personal data — it exists only so the daily analysis quota is
  metered per Agent instance instead of being shared across every user. No
  account data, no transaction history, no portfolio or credential information
  is collected or transmitted.
• All analysis runs on the publisher's backend (rocketstock.zeabur.app),
  which sources public market data from yfinance, Financial Modeling Prep,
  SEC EDGAR, Morningstar, Tavily, FinMind, and Taiwan's MOPS.
• Reports are generated per request; the Agent does not build a profile of you.