/company-at-a-glance

name: company-at-a-glance
description: >
  Produce a rigorous, fully-sourced data snapshot of a single company - a detailed multi-year financial
  table (revenue, gross profit, EBITDA, EBIT, net income per year), computed margins and growth with every
  derived figure footnoted to its formula and inputs, plus web-sourced profile facts (founded, CEO,
  ownership, employees, listing). Use whenever the user wants a company "at a glance", a snapshot / fact
  sheet / profile, "pull the financials for X", "what do we know about X", "give me the numbers on X", or
  names a single company to size up - even if they don't say "at a glance". Financials come ONLY from the
  Economic Mind financials connector and are flagged reported-vs-not-reported; profile facts are
  web-searched from credible sources and individually cited. The deliverable is data (Markdown + JSON);
  presentation is left to the client.

Company at a glance

Produce a one-page, **fully-sourced** snapshot of a single company. The deliverable is **data** - a Markdown report plus a JSON sidecar - not a styled document; how it is laid out is the client's job. The entire value is **trust**: every number is either reported (with its provider named) or shown as `n/a`, every derived figure carries a footnote with its formula and the exact inputs, and every profile fact is attributed to the source it came from. A reader should be able to audit any figure in the snapshot back to its origin without taking anything on faith.

Data sources (no exceptions)

There are **two source channels, and they are never mixed**. Keeping them separate is what lets a reader trust the numbers: the financials stay pristine and machine-traceable; the qualitative colour is clearly labelled as such.

1. **Financials & figures - ONLY the Economic Mind connector.** Every revenue, profit, margin and growth figure must trace to a `get_company_financials` call. Do **not** pull a financial number from the web, from memory, from a market report, or from a "sanity check" - even if the connector is slow, returns nothing, or seems wrong. If Economic Mind does not report a figure, it is `n/a`. There is no fallback. 2. **Profile facts - web search, credible sources only.** Founded, CEO, ownership, employees, listing status and a short "about" are qualitative context. Web-search them (see Phase 3), cite each one individually, and **never let a web-sourced fact enter a calculation.** If you cannot source a fact from a credible source, it is `n/a` - do not guess.

Work to files first, then read them back

The connector response - not your memory of it - is the single source of truth, because context can be compressed mid-task and a half-remembered number is exactly how errors creep in. So: **immediately after each tool call returns, write its raw output to a file**, before making the next call. Then, before you assemble the snapshot, **read every file back** and build the report from the files, not from recall. Use a working directory you create once at the start, e.g. `em-glance/`.

Calculations are explicit, gated, and verified

You compute the derived figures yourself - there is no bundled script - so the discipline has to come from method, not from a black box:

• **Gate on reported inputs.** Compute a figure **only when every input it needs is reported** for the

relevant year(s). If EBIT is not reported for FY2024, EBIT margin for FY2024 is `n/a` - never infer it.

• **Show the work.** Every derived figure gets a footnote with the formula and the actual numbers

substituted in, e.g. `EBITDA margin (FY2024) = EBITDA ÷ Revenue = 19.4 ÷ 121.6 = 15.9%`. A figure a reader cannot recompute from its footnote does not belong in the snapshot.

• **Verify before you use.** For any multi-step figure (e.g. a CAGR), write the steps out and check the

intermediate result before relying on it. When in doubt, recompute from the raw file rather than reusing a number you calculated earlier.

Calculation methods (use these exact definitions)

All margins are expressed as % of revenue for the same fiscal year:

• `Gross margin (FYn) = Gross profit ÷ Revenue`
• `EBITDA margin (FYn) = EBITDA ÷ Revenue`
• `EBIT margin (FYn) = EBIT ÷ Revenue`
• `Net margin (FYn) = Net income ÷ Revenue`

Growth (only across years where revenue is reported at both endpoints):

• `Revenue YoY (FYn) = (Revenue_FYn − Revenue_FYn-1) ÷ Revenue_FYn-1`
• `Revenue CAGR, full span = (Revenue_latest ÷ Revenue_earliest)^(1 ÷ (Y−1)) − 1`, where `Y` = number of

reported revenue years in the span.

• `Revenue CAGR, L3Y = (Revenue_FYn ÷ Revenue_FYn-3)^(1 ÷ 3) − 1` (needs four consecutive reported years).

State the exact years used in every growth footnote, because spans differ when filings are incomplete.

Source attribution: name the underlying provider

Economic Mind aggregates filings from underlying providers, and each line item carries its own `source` (e.g. "Proff (NO)", "S&P Capital IQ"). Name the **underlying provider** and say it was gathered through Economic Mind - e.g. *"Company filings via Proff (NO), gathered through Economic Mind"* - rather than crediting "Economic Mind" as if it were the primary source. This mirrors how the figure is actually sourced and lets the reader trace it.

Workflow

Phase 1 - Resolve the company & set up

1. Create the working directory once: `mkdir -p em-glance`. 2. `search_companies(["<name>"])`. Pick the entity matching the user's intent, using `headquarters` and `summary` to disambiguate. If two are plausible, say which you chose and why; if genuinely ambiguous, ask before continuing. 3. Write `em-glance/company-info.txt` with: `companyId`, `companyName`, `headquarters`, the EM `summary`, and the currency you will use.

Phase 2 - Pull the financials (one currency, five years)

1. Request the **five most recent years** including the latest: `get_company_financials([id], years=[2021,2022,2023,2024,2025], currency=<one