RANDOM.ORG
Utilities & PersonalGenerate true random numbers from atmospheric noise through the RANDOM.ORG API, rather than the pseudo-random generators built into code. Agents run fair draws, sampling, and audit selection that will survive scrutiny.
What This Integration Enables
Every programming language ships a random function, and for shuffling a playlist it is fine. It stops being fine the moment somebody has standing to argue with the result. A pseudo-random generator produces a sequence from a seed, which means the sequence is reproducible by anyone who knows the seed, and the fairness of the draw rests entirely on your word that nobody did. RANDOM.ORG derives its values from atmospheric noise, a physical entropy source with no seed to know. That distinction is the entire product, and it only matters for the class of work where the output has to survive an outside challenge: prize draws, audit sample selection, statistical sampling, tie-breaking, and anything a regulator might ask you to explain a year later.
The Signed API is where the argument actually gets won. A signed generation returns the values together with a cryptographic signature, a serial number and a license record, so an entrant, an auditor or a regulator can confirm the values came from RANDOM.ORG and were never altered, without needing to trust your logs. Signed results are retained for at least 24 hours and can be re-fetched by serial number, so an audit does not depend on your archiving discipline. Tickets go one step further and add commit-and-reveal: the ticket is created before the draw, consumed by the generation, and only later revealed, which is what separates a result you committed to in advance from a result you announced after seeing it. The basic operations are all here too, from integers and Gaussians through to strings, UUIDs and binary blobs for seed material.
Without FlowRunner
With FlowRunner
Use Case Scenarios
A prize draw that can be argued with and still stand
A promotion closes with several thousand entries held in Airtable. An agent deduplicates the list, records the final count, and creates a ticket before generating anything, which fixes the commitment in time. It then calls Generate Signed Integers with replacement disabled so no entrant can be drawn twice, and runs Verify Signature on the response before treating it as real. The winning positions exist, sealed, with a serial number attached. Nothing is announced. When an entrant later disputes the outcome, the answer is not a description of the process; it is a serial number they can check themselves.
Audit sample selection that holds up in the review
An internal audit team samples transactions for testing each quarter, and the selection method is itself part of what gets reviewed. An agent pulls the population from NetSuite, calls Generate Signed Integers to pick the sample positions, and stores the serial number alongside the working papers rather than only the selected rows. Months later, when the reviewer asks how the sample was chosen, Get Signed Result re-fetches the original random object and signature by serial number. The selection is reproducible as evidence without the team having had to archive anything themselves.
Issuing codes without exhausting the quota mid-run
A campaign needs a large batch of unguessable voucher codes. An agent calls Get API Key Usage first and compares the estimated remaining bits and requests against the size of the job, because a batch that dies partway leaves an incomplete code set and an unclear record of which codes were actually issued. If the quota covers the run, the agent calls Generate Strings with replacement disabled so every code is unique, and loads them into Shopify. If the quota is short, it reports the shortfall and the number of codes it can safely produce, and waits for a person to decide whether to split the run or raise the quota.
Human-in-Loop Highlight
The one-way door in this connector is Reveal Tickets. Reveal marks the named ticket and every predecessor in its chain as revealed, so the values behind them become disclosable, and the operation has no counterpart that puts the seal back. It is also the exact moment a draw stops being an internal result and becomes a public one, which is precisely when an ineligible winner, a duplicated entry or a stale entrant list turns into a communications problem rather than a data problem. So the agent runs everything up to that line and stops. It posts to the promotions owner in Slack with the facts a person needs to make the call: "Draw committed under ticket chain 4 of 4, serial 118372, signature verified. 6,412 unique entrants, three winners selected, none of them staff accounts. Reveal will publish the chain and cannot be undone. Confirm to reveal, or hold while eligibility is checked." Every earlier step, the deduplication, the generation, the signature check, runs without anyone watching. The digital andon cord is pulled at the single step that cannot be walked back, which is what human-in-the-loop is supposed to cost you: one decision, not a review of the whole run.
Agent Capabilities
21 actionsBasic Generation
7- Generate Integers Generates true random integers within an inclusive range, between 1 and 10,000 values per call with bounds anywhere in the supported interval. Disable replacement to draw unique values, which is what a shuffle or a raffle needs, or leave it enabled for independent picks such as dice rolls. Values are returned as base 10 numbers and as zero-padded strings in base 2, 8 or 16.
- Generate Integer Sequences Generates up to 1,000 sequences of true random integers, each returned as its own array. Uniform sequences share one length, range, replacement setting and base. For multiform sequences, supply the per-sequence override arrays, each holding exactly one entry per sequence. The combined length across all sequences may not exceed 10,000 values.
- Generate Decimal Fractions Generates true random decimal fractions from a uniform distribution across the interval from zero up to but excluding one, between 1 and 10,000 values at a precision of 1 to 14 decimal places. Use it for Monte Carlo simulation, weighted sampling and probability thresholds where a pseudo-random generator would not be accepted.
- Generate Gaussians Generates true random numbers drawn from a normal distribution using a Box-Muller transform over uniform true randomness. Specify the mean, the standard deviation and the number of significant digits, from 2 to 14. Suited to statistical modelling and to generating test data that behaves like real data rather than like uniform noise.
- Generate Strings Generates true random strings of equal length from a character set you supply, between 1 and 10,000 strings of 1 to 32 characters each, from an alphabet of up to 128 characters. Disable replacement to guarantee every returned string is unique, which is the setting that matters for coupon codes, voucher numbers and temporary passwords.
- Generate UUIDs Generates version 4 UUIDs in accordance with RFC 4122, backed by true randomness rather than a pseudo-random generator, between 1 and 1,000 identifiers per call. Use it where an identifier must be unguessable as well as unique, such as an invitation token or an externally visible record identifier.
- Generate Blobs Generates Binary Large OBjects of true random data, encoded as Base64 or hexadecimal. Request between 1 and 100 blobs, each sized in bits and divisible by 8, with a combined ceiling of 1,048,576 bits. Typically used as cryptographic seed material, salts or nonces.
Signed Generation
7- Generate Signed Integers Generates true random integers together with a cryptographic signature proving the values came from RANDOM.ORG and were not altered. The response carries the signed random object, a base64 signature and a serial number for later retrieval. This is the operation to use for prize draws, audit selection and any process whose fairness has to be provable to someone outside your team.
- Generate Signed Integer Sequences Generates up to 1,000 cryptographically signed sequences of true random integers, each returned as its own array. Uniform sequences share one length, range, replacement setting and base, and multiform sequences take the per-sequence override arrays. The combined length across all sequences may not exceed 10,000 values.
- Generate Signed Decimal Fractions Generates cryptographically signed true random decimal fractions from a uniform distribution, between 1 and 10,000 values at a precision of 1 to 14 decimal places. The signed response can be checked later with Verify Signature to prove the values behind a sampling decision were never edited.
- Generate Signed Gaussians Generates cryptographically signed true random numbers from a normal distribution using a Box-Muller transform. Specify the mean, the standard deviation and 2 to 14 significant digits. The signature allows the sample to be audited after the fact rather than taken on trust.
- Generate Signed Strings Generates cryptographically signed true random strings of equal length from a character set you supply, between 1 and 10,000 strings of 1 to 32 characters each. Suited to verifiable prize codes and to voucher issuance that has to be reconciled against an external record.
- Generate Signed UUIDs Generates cryptographically signed version 4 UUIDs in accordance with RFC 4122, between 1 and 1,000 per call. The signature and serial number let a third party confirm the identifiers were issued by RANDOM.ORG at the stated time, which matters when the identifier itself is the entitlement.
- Generate Signed Blobs Generates cryptographically signed Binary Large OBjects of true random data, encoded as Base64 or hexadecimal, between 1 and 100 blobs within the same combined size ceiling as the unsigned operation. Use it where random seed material has to be demonstrably unmodified.
Verification
2- Verify Signature Verifies that a signed response really originated from RANDOM.ORG and has not been altered. Pass the complete random object and the signature string exactly as they were returned. The authenticity flag comes back true only when every field of the random object still matches the signature, which makes this the right validation step before an agent acts on a signed result.
- Get Signed Result Retrieves a previously generated signed result by its serial number. Signed results are stored for at least 24 hours, so an audit can re-fetch the original random object and signature without the caller having archived them. Quota fields in the response are historical and should not be used to drive behavior.
Tickets
4- Create Tickets Creates between 1 and 50 tickets, each consumable once by a signed generation operation, producing a tamper-evident audit trail. The Show Result setting controls whether Get Ticket later discloses the values the ticket was used for, which is what makes a commit-and-reveal draw possible: the outcome exists but stays hidden until reveal time.
- Get Ticket Returns the details of a single ticket, including its hashed API key, creation and usage timestamps, expiration time, chain neighbors, and the serial number of the result it was used for. When the ticket was created with Show Result enabled, or has since been revealed, the full random and signature objects are included.
- List Tickets Lists tickets belonging to the configured API key, filtered by position in a ticket chain: singletons stand alone, chain heads start a chain and chain tails end one. It returns up to 4,000 tickets with their timestamps, serial numbers and chain neighbors, which is how an agent reconciles a series of draws after the fact.
- Reveal Tickets Marks a ticket and every predecessor in its chain as revealed, so Get Ticket returns the full random and signature objects for them. This is the reveal step of a commit-and-reveal draw and it does not run in reverse. Successors are unaffected, and revealing an already revealed chain returns a count of zero.
Account
1- Get API Key Usage Returns quota and status information for the configured API key, including whether the key is running or stopped, when it was created, the estimated bits and requests remaining, and the lifetime bits and requests consumed. Use it as the guard step before a large generation job, or as the source for an alert when a key is close to exhausted.
Frequently Asked Questions
What can FlowRunner do with RANDOM.ORG?
FlowRunner agents can run Generate Integers, Generate Integer Sequences, and Generate Decimal Fractions in RANDOM.ORG, plus 18 more actions.
Does connecting RANDOM.ORG to FlowRunner require OAuth?
No. RANDOM.ORG connects to FlowRunner with an API key, no OAuth flow required.
Can RANDOM.ORG trigger a FlowRunner workflow automatically?
RANDOM.ORG doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with RANDOM.ORG
$100 in credits. No card required. Connect in minutes.