Skip to content

diffbot/diffbot-skills / skills/diffbot-deals

diffbot-deals

Unverified

Funding rounds, investments, and acquisitions in the Diffbot Knowledge Graph by company, investor, acquirer, round series, amount, date, or the company's industry. MUST USE skill when the answer involves a list of investment deals: deal flow, a company's funding history, an investor's portfolio activity, M&A lists, or the largest deals in a period. Rows are investment deals. Triggers on: funding rounds, raised, Series A, venture funding, funding history, deal flow, deal size, investments in, acquisitions, M&A, acquired for, valuation round.

Diffbot Deal Search

Find funding rounds, acquisitions, and transactions in the Diffbot Knowledge Graph. Deals live in two places — as standalone Investment/Transaction records, and as fields hanging off the Organization involved. Choosing the right one is the whole skill; see the routing table below.

First, check the row shape. This skill is for when the rows are deals — target, acquirer/investors, amount, date, series. If the rows the user wants are companies ("everything Microsoft acquired", "companies Sequoia backed"), hand off to /diffbot-organizations; it filters on the same acquiredBy and investments fields but presents companies.

Sibling skills: /diffbot-organizations (companies as rows), /diffbot-news, /diffbot-places. Use /diffbot-dql for anything outside these shapes.

Workflow

Step 1 — check the CLI, then db dql init

~/.diffbot/venv/bin/db --version
~/.diffbot/venv/bin/db dql init

The db CLI is a one-time user install at ~/.diffbot/venv/bin/db, the same way gh or wrangler is — this skill never installs, upgrades, or uninstalls it. If --version fails with No such file or directory, stop and show the user this line to run or approve themselves (it is deliberately not pre-approved):

python3 -m venv ~/.diffbot/venv && ~/.diffbot/venv/bin/pip install 'diffbot>=3.1.1'

If it prints a version below 3.1.1, fails with No such option '--version' (an old diffbot-python install, which predates the flag), or warns about the diffbot-python rename, point the user at Upgrading the db CLI in the plugin README and stop. Never run an install, upgrade, or uninstall line yourself.

Save output files under .diffbot/tmp/ in the working directory. db creates that folder itself, so don't mkdir it first. db, jq, and cat are not pre-approved — Claude will ask before running them.

init refreshes ~/.diffbot/ontology.json and verifies a token (DIFFBOT_API_TOKEN env var, else ~/.diffbot/credentials). If neither exists the user must run:

echo "DIFFBOT_API_TOKEN=YOUR_TOKEN_HERE" > ~/.diffbot/credentials && chmod 600 ~/.diffbot/credentials

Tokens: https://app.diffbot.com/get-started/. Never echo the token.

Step 2 — route the question

The user asks aboutQueryWhy
Funding rounds by date, size, series, or investortype:InvestmentOne row per round, with amount, series, date, investors
Deals in an industry / location / company-size segmenttype:Organization + investments.{...}Industry lives on the company, not the deal — see the gap below
Acquisitions ("who bought X", "everything Microsoft acquired")type:Organization isAcquired:true acquiredBy.name:"…"Acquisition records are name-only stubs
Any money movement, not just equitytype:TransactionSuperset: Investment and Acquisition are both subtypes

The industry gap — read this before filtering deals by sector

investee on an Investment is a LinkedEntity, which carries only name, types, summary, image, diffbotUri. It has no categories, so investee.categories.name:"…" returns zero — it looks like "no such deals" rather than "no such field."

Flip the query to the company side instead, and use {} so the round's conditions co-constrain:

type:Organization categories.name:"Artificial Intelligence Software" investments.{amount.value>50000000 date>="2026-01-01"}
type:Organization categories.name:"Artificial Intelligence Software" investments.{series:"Series A" date>="2026-01-01"}

Without {} the clauses are independent — a company with any $50M round and any 2026 round matches, even if they're different rounds.

Step 3 — field reference

type:Investment — a funding round.

FieldTypeNotes
investeeLinkedEntity (Organization)investee.name — the company raising
investment.seriesString"Series A", "Series B", "Series Unknown", …
investment.amount.valueFloatRaw number; pair with .currency
investment.amount.currencyString"USD", … — always filter or display currency, values are not normalized
investment.dateDDateinvestment.date>="2026-01-01"
investment.investorslist of LinkedEntityinvestment.investors.name
date, name, amountInherited from Transaction; date mirrors investment.date

type:Transaction — any transaction; Investment and Acquisition are subtypes (filter with types:"Acquisition").

FieldNotes
payeeLinkedEntity (Organization) — receiving side
payerslist of LinkedEntity — paying side
amount.value / amount.currency
dateDDate
namee.g. "Venture Round - OpenAI"

Organization.acquiredBy — the reliable path for M&A.

FieldNotes
acquiredBy.nameThe acquirer
acquiredBy.amount.valueDeal size
acquiredBy.date
isAcquiredBoolean flag on the target

Also on Organization: investments (full round history), totalInvestment.value, nbUniqueInvestors.

Sorting. Deals default to newest first: append revSortBy:investment.date on type:Investment, revSortBy:investments.date when the query runs on the Organization side, and sortBy:date on type:Transaction. Unsorted, the ranking surfaces the headline rounds instead (investment.investors.name:"Sequoia Capital" → OpenAI $122B, Anthropic $50B, OpenAI $40B), which answers "biggest" rather than "recent"; when size is the question, say so with:

revSortBy:investment.amount.value # largest rounds first

This is the one deliberate exception, alongside diffbot-news, to the no-sort rule the other skills follow, and it matches diffbot-mcp's search_deals default.

Worked examples

type:Investment investment.investors.name:"Sequoia Capital" revSortBy:investment.date
type:Investment investment.series:"Series A" investment.date>="2026-01-01" revSortBy:investment.amount.value
type:Investment investment.amount.value>100000000 investment.amount.currency:"USD" revSortBy:investment.date
type:Investment investee.name:"OpenAI" revSortBy:investment.date
type:Organization isAcquired:true acquiredBy.name:"Microsoft" revSortBy:acquiredBy.amount.value
type:Organization categories.name:"Semiconductor Companies" investments.{amount.value>100000000 date>="2025-01-01"} revSortBy:investments.date
type:Transaction types:"Acquisition" sortBy:date

Note on type:Acquisition. These records exist (~209k) but are near-empty stubs — typically just a name like "Reliance Motor Car Company acquired by General Motors", with date, amount, payee, and payers all null. Don't build a deal table from them; route acquisitions through Organization.acquiredBy.

Distribution questions are facets, not row lists

"What stage is most funding at?", "which investors are most active?" — aggregate instead of listing:

~/.diffbot/venv/bin/db dql export 'type:Investment investment.date>="2026-01-01" facet:investment.series' \
  --out .diffbot/tmp/facet.json --format json --size 12

Returns buckets like Seed 4568 · Grant 3796 · Series A 2061 · Debt Financing 2015 · Pre Seed 1965. Also useful: facet:investment.investors.name (most active investors), facet:investment.amount.value (auto-bucketed deal sizes).

Facets need --size ≥ 1 — --size 0 errors — and --size sets the number of buckets, not rows. A facet response has value/count per bucket, not entities, so don't use it when the user wants individual deals.

or() is usually unnecessary on series. investment.series:"Series A" is a contains match and already covers "Series A-1"; or("Series A","Series A-1") returns the identical count. Reach for or() only across genuinely different strings.

Step 4 — probe before committing

~/.diffbot/venv/bin/db dql probe \
  'type:Investment investment.series:"Series A" investment.date>="2026-01-01"' \
  'type:Investment investment.amount.value>100000000' \
  'type:Organization categories.name:"Artificial Intelligence Software" investments.{amount.value>50000000 date>="2026-01-01"}'

Zero hits on an investee.<something> path almost always means the field doesn't exist on LinkedEntity — re-read the gap above rather than loosening the filter. probe fails the whole batch if any one variant is rejected by the API, so fix the bad clause and re-run.

Step 5 — export and display

Rounds:

~/.diffbot/venv/bin/db dql export "<DQL>" --out .diffbot/tmp/deals.csv \
  --spec "investee.name,Company;investment.series,Round;investment.amount.value,Amount;investment.amount.currency,Currency;investment.date.str,Date;investment.investors.name,Lead Investor" \
  --size 50

Acquisitions:

~/.diffbot/venv/bin/db dql export "<DQL>" --out .diffbot/tmp/acquisitions.csv \
  --spec "name,Target;acquiredBy.name,Acquirer;acquiredBy.amount.value,Amount;acquiredBy.date.str,Date;categories.name,Industry" \
  --size 50

Export JSON when the full investor list matters — --spec renders only the primary value of a list field, so the CSV shows one investor even when a round had thirty:

~/.diffbot/venv/bin/db dql export "<DQL>" --out .diffbot/tmp/deals.json --format json --size 50
jq -r '.data[].entity | "\(.investment.date.str)\t\(.investee.name)\t\(.investment.series)\t\([.investment.investors[]?.name] | join(", "))"' .diffbot/tmp/deals.json

Display

  1. Render a markdown table: Date | Company | Round | Amount | Investors.
  2. Reformat amounts. CSV emits scientific notation (3.0E8) — show $300M. Never present a bare number without its currency.
  3. Dates come back as d2026-07-23 — strip the leading d.
  4. Blank amount means the round was reported without a disclosed size — label it "undisclosed", not 0.
  5. "Series Unknown" is a real value in the data, not a lookup failure.
  6. Print the final DQL in a plain code block, give the saved file path, and offer more rows (--size N, --from K) or a refinement.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 467kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 319kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 319kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 319kUpdated Claude CodeCodexVerifiedInstall