Finterm.ai：为Claude Code打造的Bloomberg终端式金融数据工具

§ 01 · Features

Finance tools shaped for agent context.

Finterm gives agents both the raw financial data and the token-efficient layer on top: filtered sources, calculated signals, diffs, and structured outputs.

Ticker data bundle

finterm bundle run ticker_data NVDA --delivery-mode inline_result

Creates the NVIDIA ticker_data run and returns the structured ticker snapshot inline.

The full ticker snapshot in one bundle: earnings, guidance, price reaction, ratios, options sentiment, short pressure, technicals, statements, and pre-earnings context.

EarningsGuidancePricesRatiosOptionsStatements ticker_data.NVDA.yaml inline result Earnings actuals + surprise Guidance revenue + EPS Prices T-1 / T0 / T+5 Options P/C + liquidity Technicals RSI · MACD · SMA ticker_data bundle · live/current output

SEC filing diffs

finterm tool sec_filing_diff NVDA --base 2025:Q3 --compare 2026:Q1

Compares NVIDIA filings section by section and returns the changed language.

Item 7. MD&A — First Quarter FY2027 Summary − Revenue $57.0B · GM 73.4% · net income $31.9B · EPS $1.30 (Q3 FY2026) + Revenue $81.6B · GM 74.9% · net income $58.3B · EPS $2.39 (Q1 FY2027, +85% YoY) Item 1A. Risk Factors — China export controls − ≈$50M H20 revenue under USG licenses to China-based customers (Aug 2025) + H200 now carries a 25% U.S. import tariff; no Hopper shipments to China this quarter vs $4.6B a year ago + New: $18.6B into private cos/infra funds; up to $10B in Anthropic; equity stake in Intel pending approval MD&A · 131% churnRisk Factors · 60% churnQA-checked

Ticker Deep Research

finterm bundle run company_deep_research NVDA --company-name "NVIDIA Corporation" --param q=Q1 --param fy=2027 --param prev_q=Q3 --param prev_fy=2026 --delivery-mode summary_json

Runs the search, crawl, dedupe, and source-labeling pass for NVDA.

Thorough by design: fetches 600–800 links per ticker, then strips the 30–40% that is noise—AI slop, SEO spam, syndicated reprints—leaving a comprehensive read of what the market thinks about the stock.

742 crawled → 0 kept Strip AI slopDedupe syndicationLabel primary / secondary 40 searches · 742 sources crawled · 272 dropped as noise, dupes, and off-topic Crawled · 742 NVDA options flow & screener (fintel.io · template) Nvidia demand continues to outweigh supply, JPMorgan says (finance.yahoo.com) NVIDIA price & institutional ownership (fintel.io · boilerplate) iFlow: earnings-season outlook shifts (bny.com · off-entity) NVIDIA Q4/FY2026 financial results (nvidianews.nvidia.com) Nvidia halts H200 output as China export controls bite (finance.via.news) Q4 results press release — reprinted across 11 sites merged Nvidia Q4 call — WATCH LIVE (youtube.com · low-signal) Analysts keep hiking targets — is NVDA too cheap? (barchart.com) Q4 2026 earnings call transcript (marketscreener.com) 2026 market & economic outlooks (morganstanley.com · off-topic) 85% GPU share faces competition; DOJ probe (finance.yahoo.com) Kept · 470 Nvidia demand continues to outweigh supply, JPMorgan says (finance.yahoo.com) Secondary NVIDIA Q4/FY2026 financial results (nvidianews.nvidia.com) Primary Nvidia halts H200 output as China export controls bite (finance.via.news) Secondary Analysts keep hiking targets — is NVDA too cheap? (barchart.com) Secondary Q4 2026 earnings call transcript (marketscreener.com) Primary 85% GPU share faces competition; DOJ probe (finance.yahoo.com) Secondary + 464 more sources kept, deduped and labeled 0 AI slop dropped 0 duplicates merged 0 kept and labeled

Options sentiment

finterm tool options_overview NVDA

Reads the live options chain into put/call flow, IV30, and expected moves.

Neutral-bullish Call-tilted flow · IV30 below realized vol Puts Calls 0% call share of volume P/C (vol) 0.72 IV30 36.9% Exp. move (1mo) ±8.7%

Insider and 13F flows, netted

finterm tool insider_trades NVDA

Returns NVIDIA Form 4 insider transactions with the trailing 90-day open-market net.

SourceFilerActionSharesValue Form 4 M. Stevens · Director Sell −1.00M −$221.1M Form 4 A. Puri · EVP Ops Sell −270,134 −$49.2M Form 4 C. Kress · CFO Sell −30,201 −$5.3M Form 4 S. Gawel · PAO Grant +59,509 RSU 13F Rathbones Group Hold 6.05M $1.05B Open-market Form 4 · NVDA · 90d $0.0M net insider selling

Ticker sentiment

finterm tool ticker_sentiment NVDA

Scores seven components against NVDA’s own trailing year and returns the 0–100 composite.

0 extreme fear up from 6 a week ago · Δ5d +8.2 coverage 5/7 Fear Greed Trend 10 Flow — Positioning 20 Price vs 125-day average 17 52-week range position 2 Up-day volume share (20d) 12 Short interest 13 Implied-volatility rank 27 Put/call volume (5-day avg) omitted · history accruing Open-interest put/call tilt omitted · history accruing Each component scored vs NVDA’s own year · 100 = greed side § 02 · Why it exists

Finterm connects the data first, so your agent can reason next.

It turns the messy inputs behind a trade into sourced, timestamped outputs: ticker data, filing diffs, ownership flows, market context, and deep research with the noise stripped.

finterm bundle run ticker_data NVDA One command

Give your agent a question and let the CLI pull the relevant financial context.

ticker_data.AAPL.yaml Agent-readable output

Results come back as structured data your agent can read, cite, and reuse.

SEC 8-K · 12:04Z ✓ Earnings call · Oct 30 ✓ Sources included

Facts carry source links, timestamps, and caveats instead of loose summaries.

§ 03 · Full demo

One prompt. Three filers. The pattern hiding in plain sight.

A real run: the prompt, your agent’s sourced answer, and the exact finterm calls underneath.

finterm sec_filing_diff · GOOGL · AMZN · MSFT Show all Replay ↻ can you use finterm to pull SEC filing diffs for the big cloud providers (google cloud, aws, microsoft) and find any patterns they all mention? just the last 4 quarters of 10-Qs can you use finterm to pull SEC filing diffs for the big cloud providers (google cloud, aws, microsoft) and find any patterns they all mention? just the last 4 quarters of 10-Qs

Mapped the three cloud brands to their SEC filers (Alphabet/GOOGL, Amazon/AMZN, Microsoft/MSFT) and diffed consecutive 10-Qs section by section.

Ran SEC filing diff · Alphabet (GOOGL) Bash $ finterm tool sec_filing_diff GOOGL --base 2025:Q3 --compare 2026:Q1 MD&A, Risk Factors changed · QA-checked Ran SEC filing diff · Amazon (AMZN) Bash $ finterm tool sec_filing_diff AMZN --base 2025:Q3 --compare 2026:Q1 MD&A, Risk Factors changed · QA-checked Ran SEC filing diff · Microsoft (MSFT) Bash $ finterm tool sec_filing_diff MSFT --base 2026:Q2 --compare 2026:Q3 MD&A, Risk Factors changed · QA-checked

Four themes show up across all three filings:

1. An AI-capex step-change. All three roughly doubled quarterly capex year over year, tied explicitly to AI infrastructure.

   “During the three months ended March 31, 2025 and 2026, we spent $17.2 billion and $35.7 billion on capital expenditures, respectively.” Alphabet 10-Q · FY2026 Q1 · MD&A

2. AI written into the risk factors. Each amended long-standing risk language to name AI as a new vector.

   “…building AI-optimized infrastructure, including our custom TPUs… Our investments ultimately may not be commercially viable or may not result in an adequate return of capital.” Alphabet 10-Q · FY2026 Q1 · Risk Factors

3. Multi-billion stakes in frontier-AI labs. The build-out now runs through the model labs themselves.

   “In Q1 2026, we invested $15.0 billion in OpenAI’s Series C Preferred Stock and entered into an equity commitment letter agreement to purchase an additional $35.0 billion…” Amazon 10-Q · FY2026 Q1 · MD&A

4. Off-balance-sheet data-center and power exposure. Growth leans on uncommenced leases and credit backstops for infrastructure they don’t own.

   “…up to $33.3 billion of future backstops to support the build-out of data center and energy supply infrastructure, alongside uncommenced data-center leases of $75.6 billion.” Alphabet 10-Q · FY2026 Q1 · MD&A

Net: across Google Cloud, AWS, and Azure, the last few quarters of 10-Qs converge on a capital-intensive AI build-out that is now visible in both the financials and the risk language. Every quote links to the exact filing.

Real Finterm output · figures quoted from public filings Not investment advice § 04 · Prompts to try

Point your agent at the hard questions.

The same shape: your prompt, then Finterm’s sourced answer, across the questions a serious analyst asks.

Is this dip in NVDA being bought or sold by insiders? Pull Form 4 activity and the latest 13F holders, and net it out.

Give me the bull and bear case on AMD’s data-center demand this quarter, from multiple sources, deduped and labeled primary or secondary.

What’s the options market pricing into TSLA earnings, and where does sentiment sit versus its own trailing year?

Diff Apple’s latest 10-K against last year’s and flag every new or softened risk factor.

§ 05 · Everything in the bundle

Everything in the bundle—and what it costs bought separately.

One subscription replaces four or five. The comparison, category by category—including the rows you can buy elsewhere.

In the bundle Comparable cost The company packet. Earnings, guidance, price reaction, ratios, options, short pressure, technicals—one call. $60–150/mo SEC filing diffs. Section-aware, line-by-line diff of any two filings, churn stats, QA-checked. No real equivalent SEC search and sections. Find any filer’s filings; pull clean narrative sections as text. $49/mo Ownership flows, netted. Form 4 trades rolled into a 90-day open-market net; 13F holders, ticker-mapped. $49–149/mo Options intelligence. Implied versus realized with rank, today’s flow, positioning, expected moves—one call. $99–199/mo Sentiment as a number. A 0–100 composite of trend, flow, and positioning, each scored against the ticker’s own year. $59–99/mo Ticker Deep Research. 600–800 links per ticker filtered into one research packet: noise stripped, reprints merged, sources tiered, provenance kept. No real equivalent Prices and statements. Live prices, reported financial statements, standard technicals. $20–99/mo The dataroom. Every run cached to a local, re-queryable room — produced once, reused every session. No real equivalent

Purchased separately: $300–450/month (APIs alone)

Finterm Pro: $200/month (fetched data cached, yours to keep)

Category price ranges from published provider price lists, as of July 2026.

Give your agent a finance brain.

Sign up $150/month $200/month 3-day free trial

3-day free trial, card required.

Special preview price for the first 1,000 users: $50 off for a limited time during the preview period.

Once you have an account, just give finterm to your favorite coding agent:

In your terminal npm install -g @finterm-ai/cli Copy Or paste into your coding agent

I want to set up finterm.ai: run `npm install -g @finterm-ai/cli`, then `finterm setup` to install your agent skills and `finterm prime` to orient yourself. I’ll handle the browser step of `finterm auth login` when you’re ready—then show me what Finterm can do.

Copy

Not investment advice.