Upload .docx files to chart flagged key term frequency and see suggested replacements plus how much preferred vocabulary the draft already uses. Processing is 100% in your browser; no file ever leaves your computer. How it works
100% client-side. Uploaded files are parsed in your browser; no content is sent to any server. A strict Content Security Policy is enforced, and the tool works offline via a service worker.
The default key term list builds on The New York Times' list of words flagged across federal agency websites (NYT, March 7, 2025), plus FedInt engine findings. Upload or paste your own to override it. Download a sample keywords.txt to test.
Load a key term list and one or more .docx files, then Generate.
How many of the currently-preferred key terms your draft already includes. Higher means the draft uses more of the rising vocabulary.
Key federal terms over time: how often each appears in federal documents, plus a then-versus-now split into rising, falling, and retired. Generated by the FedInt engine from federal sources.
No engine intel found at data/current/. Run fedint run … to generate artifacts, then reload.
Explore Key term momentum over time
How often each tracked word appears in Federal Register documents, month by month. Coverage starts in 1994, the earliest Federal Register data available in machine-readable form, and runs to now. It opens on all time; change the From/To range or pick an administration to focus. Choose which key terms to plot, or check rows in the table. A rising line means the government is using that word more; a falling line means less.
Momentum
Window B
| Key term | Start | Now | Δ |
|---|
Then vs now What the government started and stopped saying
This compares an earlier stretch of federal documents against a recent one and sorts the vocabulary three ways. Rising: words appearing much more now. Falling: words appearing less often. Retired: words that were common before and have all but vanished. It shows which language the government currently favors and which has faded.
How many key terms moved each way
Themes Named clusters
Related key terms grouped into themes, so you can see the shift as topics (for example "AI & Innovation" or "Governance & Stewardship") rather than one word at a time.
Key terms per theme
Dominant key terms For the selected timeframe
The rising vocabulary spelled out. The words appearing most in current federal documents, with why each one recurs and which agencies use it.
For every flagged word, the safer alternatives you can use instead. The list comes from The New York Times' reporting on words drawing federal scrutiny.
Suggested alternatives Fading words and rising alternatives
Words that are fading from federal documents, each paired with the rising words you could use instead. These pairings are drawn from frequency alone; a full analysis run produces evidence-backed, curated matches.
Every flagged word and its safer alternatives
The full New York Times list of words drawing federal scrutiny, each paired with substitutes you can use instead.
Each award-making agency's tracked vocabulary, split into flagged terms (the NYT-flagged words drawing scrutiny) and safer alternatives to use instead. Pick a window; an up arrow means the agency used the term more toward the end of it, a down arrow less. A ⊘ on a flagged term means most of its mentions are ban language, so a rising arrow there is rising prohibition, not adoption; a ~ marks a mixed signal. A marks a term on the NYT's published list (the rest were surfaced by the engine). Frequency-derived from federal funding and rulemaking text, so smaller agencies show thinner signal.
No agency data found. Run fedint backfill-agency to generate it.
The primary sources, data feeds, and prior work behind FedInt's intelligence. Where the dashboard cites a rule by section and page, the source is here. Regulations are the ground truth; news is a discovery layer, never a frequency input.
References list unavailable.
Funding themes across the signal stack, earliest to latest: RHET (Rhetoric — Federal Register policy language; leads ~1–2 years), APPR (Appropriated — NIH RCDC spending intent with a forward estimate; leads awards 6–18 months), and the Awarded badge (obligated dollars; trailing). A theme rising in rhetoric and intent but flat in awards is the earliest opportunity. Click a row to drill into the Institutes & Centers, the three signals, and the framing language driving it.
Awarded $ from obligations only. APPR = NIH RCDC theme-level funding intent (USD, fiscal-year, latest actual + estimate), shown only for themes with a clean RCDC category. RHET = Federal Register document mentions per year for the theme's policy query, shown only for policy-shaped themes (funded-not-regulated themes carry no rhetoric). Each badge hover shows the figures behind it.
Every NIH research priority the government funds, ranked by momentum. Subscribe to the ones you track (free: up to 5) and the banner tells you which moved since the last update. Awarded $ is the obligated-spending signal (RCDC, official); RHET is Federal Register policy language (leading indicator). Source labels distinguish official figures from estimates.
Awarded = NIH RCDC obligated spending (USD, fiscal-year, actual years only). RHET = Federal Register document mentions per year (featured categories + the top funded; others carry awarded-only). "Moved" = a category whose awarded or rhetoric momentum changed since the prior snapshot. Subscriptions are stored in this browser only.