Spectra - Complete Guide
Spectra is a browser-based crypto investigation platform combining blockchain forensics with open-source intelligence.
Introduction
Spectra is an investigation platform that runs entirely in your browser. It brings together two disciplines that usually live in separate tools: on-chain tracing — following value across hops, through instant exchanges and around mixers — and open-source intelligence — the work of connecting an address to a person.
There are twelve modules. They share one evidence model, so a finding in one is usable in the next without re-keying anything.

Contents
Getting access
Spectra lives at spectra.intelligenceonchain.com. There is no install and no API key to configure — every data source is handled server-side.
There are two routes in.
Three tiers, each a monthly credit balance plus a set of unlocked modules.
Basic
30
8 of 12
Pro
75
10 of 12
Enterprise
400
12 of 12
Credits reset to the full balance on your renewal date rather than rolling over. Upgrading, downgrading or resubscribing also triggers a refresh.
You can upgrade your subscription, using the 'Subscription' button at the top right of the screen


Holding 100 IOC token unlocks Spectra without a monthly subscription.
Sign in with the wallet holding your balance and access is verified on-chain. You do not transfer or lock the tokens — the check is a read of your balance at sign-in.
Check the current threshold before buying, as it is set by Intelligence On Chain and can change.
What each tier unlocks
Domain Search
✅
✅
✅
Wallet Mentions
✅
✅
✅
Username Deep Search
✅
✅
✅
Obsidian Case Manager
✅
✅
✅
Dork Trace
✅
✅
✅
Domain Quick OSINT
✅
✅
✅
Email Extractor
✅
✅
✅
Exchange X-ray
✅
✅
✅
Temporal Analysis
—
✅
✅
Multi-Wallet Researcher
—
✅
✅
Tornado Withdrawal Viewer
—
—
✅
Tornado Cluster Investigator
—
—
✅
How credits work
One credit is the unit of work. Every module run costs a fixed number, deducted when the run completes.
Domain Search
1
—
Wallet Mentions
2
5
Username Deep Search
2
—
Obsidian Case Manager
2
—
Dork Trace
2
—
Domain Quick OSINT
3
—
Email Extractor
3
—
Temporal Analysis
3
—
Exchange X-ray
5
—
Multi-Wallet Researcher
5
10
Tornado Withdrawal Viewer
20
—
Tornado Cluster Investigator
25
—
Spending sensibly
A few habits stretch a balance a long way.
Triage cheap, then commit. Domain Search costs 1 credit and Wallet Mentions costs 2. Run those first across a list of candidates, and spend the 20–25 credit modules only on the two or three that survive.
Batch the researcher. Multi-Wallet Researcher charges per wallet, so a five-wallet batch costs the same as five single runs — but takes one pass and produces one comparable report.
Deep mode is a decision, not a default. Deep search on Wallet Mentions costs 5 rather than 2. Use it when the standard run returned something worth expanding, not as the opening move.
A failed run does not charge you. Charging happens per completed unit of work, so a batch that fails partway bills only for what finished.
Your first investigation
Take an address you already know something about — a wallet from a case you closed, an exchange hot wallet, or a public entity. Knowing the answer in advance is the fastest way to calibrate what the output means.
Run Wallet Mentions (2 credits)
The cheapest useful first look. It searches the address across Google, Brave, Arkham and six social platforms simultaneously, and tells you whether it has ever been posted publicly.
If it returns nothing, that is information: an address with no web footprint is unlikely to be an exchange, a service, or a known scammer.
Follow what you found
The result determines the next move rather than a fixed sequence.
Named entity or exchange → the trail may end here; consider whether a subpoena is the next step rather than more analysis.
A username or handle → Username Deep Search.
A domain → Domain Quick OSINT.
Nothing at all → Exchange X-ray if funds moved through a swap service, or Temporal Analysis to profile the operator.
The modules
Tornado Cash analysis
This is the part of Spectra most likely to be misunderstood, so it is worth being precise.
What cannot be done
Tornado Cash severs the on-chain link between a deposit and a withdrawal. That is the entire function of the protocol and it works. No tool recovers that specific link, Spectra included. Anyone claiming to reverse the cryptography is either mistaken or selling something.
What can be done
The pool is opaque. Everything around it is not.
Recipients that transact with each other. A wallet that withdraws from a mixer and then sends funds directly to another withdrawal address has undone much of the benefit. This is the strongest available signal.
Shared intermediaries. Two recipients that both interact with the same third address are linked, more weakly, but meaningfully once other signals agree.
Withdrawal timing. Withdrawals inside a narrow window suggest one operator working through a list. On a quiet pool this is a strong signal; on a busy pool it is close to noise, which is why Spectra shows the surrounding traffic rather than the window alone.
Funding ancestry. Where the gas came from, and whether several recipients were funded from a common source.
What that adds up to
A shortlist. Forty withdrawal addresses become perhaps three worth investigating, each with a confidence figure and the reasoning that produced it.
Suggested sequence
Tornado Withdrawal Viewer (20cr) — enumerate the recipient population for the relevant pool and period.
Tornado Cluster Investigator (25cr) — map relationships across that population.
Temporal Analysis (3cr) — profile the two or three strongest candidates individually.
Wallet Mentions (2cr) and Username Deep Search (2cr) — attempt attribution on the survivors.
Roughly 55 credits for a full pass. Within an Enterprise balance this is comfortable; on Pro the first two steps are unavailable.
Workflows
Tracing a theft to an off-ramp
Starting point: the victim's wallet and the outbound transaction.
Follow the hops manually in an explorer until funds reach a service address. Spectra does not replace this step.
Identify the service. If it is an instant exchange, run Exchange X-ray (5cr) with a tight time window.
Match the output leg on timestamp and value. Widen tolerance gradually. Note any higher-value outputs the module flags.
Continue from the output address — repeat until funds reach a KYC venue or a mixer.
At a KYC venue, analysis stops and process begins: the deposit address plus your documented trail is what an exchange compliance team needs.
At a mixer, switch to the Tornado workflow.
Typical cost: 10–20 credits across two or three swap services.
Putting a name to an address
Starting point: an address with no obvious owner.
Wallet Mentions (2cr) — has it been posted publicly anywhere?
If a handle appears → Username Deep Search (2cr) across seven platforms and the breach corpora.
If a domain appears → Domain Quick OSINT (3cr). Check the certificate for sibling domains.
If a name appears → Email Extractor (3cr) against the associated domain.
Temporal Analysis (3cr) to corroborate — does the operating window match the timezone the identity implies?
Corroboration rule: two independent sources before you treat an identity as established. A username match alone is not attribution.
Typical cost: 7–13 credits.
One phishing domain to the whole operation
Starting point: a single drainer or phishing domain.
Domain Quick OSINT (3cr). The TLS certificate is the highest-value field — campaigns routinely share one across every domain.
Dork Trace (2cr) on the distinctive strings: wallet addresses in the page source, contact handles, unusual copy.
Domain Search (1cr each) to triage the sibling domains cheaply.
Wallet Mentions (2cr) on the drainer wallet found in the page source.
Multi-Wallet Researcher (5cr per wallet) if the campaign uses a set of collection addresses.
Typical cost: 15–30 credits depending on campaign size.
Reading the output
Risk and confidence are two different numbers
This is the single most important thing to understand about Spectra's output.
Risk is how concerning the wallet appears. Confidence is how much evidence that assessment rests on.
High risk, high confidence
Multiple signals agree. Act on it.
High risk, low confidence
One strong signal, nothing corroborating. Investigate further.
Low risk, low confidence
Little was found. Not the same as "clean."
Low risk, high confidence
Substantial history, nothing concerning in it.
A fresh wallet with three transactions returns low risk, low confidence, because there is nothing to assess. Tools that return "medium risk" in this situation are converting absence of evidence into evidence of absence, and that is how people get wrongly flagged.
Source badges
Every fact carries its origin — ETHERSCAN, ARKHAM, BRAVE, HIBP, CRT.SH, DERIVED. DERIVED means Spectra calculated it rather than read it, and it is the label to scrutinise when a finding matters.
Confidence bands on links
80–100%
Strong. Direct on-chain relationships.
60–79%
Moderate. Usually shared intermediaries. Corroborate.
Below 60%
Weak. Displayed faded. A lead, nothing more.
Methodology
Spectra is built on a position worth stating explicitly, because it shapes every output.
Conclusions drawn across chain data and open sources are judgements about likelihood, not facts retrieved from a database. On-chain data is complete but anonymous. Open-source data is rich but unreliable. Anything derived from combining them is probabilistic.
Four principles follow.
Confidence is scored, not implied. A 44% link is presented as a 44% link rather than rounded into the narrative.
Competing explanations stay visible. Alternatives are recorded alongside the leading hypothesis, with the evidence that would shift weight between them. This is Analysis of Competing Hypotheses, and it exists because the failure mode in investigation is not missing evidence — it is settling on an explanation early and reading everything afterwards as support for it.
Absence of evidence is reported as absence. Nothing found means low confidence, not medium risk.
Leads are labelled as leads. Clustering and timing correlation produce candidates for verification, and the interface says so in the same place it shows the result.
Exporting and case management
Every module exports Markdown or CSV. Exports carry a timestamp, the source of each fact, and the confidence rating where one applies.
Markdown for narrative work — reports, handovers, anything a person will read. Drops straight into an Obsidian vault.
CSV for data work — address lists, withdrawal enumerations, anything you will sort or cross-reference in a spreadsheet.
A workable case structure
Two habits that repay themselves:
Never edit a raw export. Keep it as retrieved and write your analysis in a separate file. When someone asks where a figure came from six weeks later, you want the original.
Keep the dead ends. The single most wasteful thing in a long investigation is repeating a line of enquiry that was already exhausted — usually by yourself.
Troubleshooting
Glossary
Attribution — connecting an address to a real-world identity. Distinct from tracing, which follows value without naming anyone.
Cluster — a set of addresses assessed as probably under common control. A conclusion about likelihood, not ownership.
Co-timed withdrawal — withdrawals from a mixer inside a narrow window, suggesting one operator working through a list.
Demixing — commonly used to mean recovering deposit-to-withdrawal links. In practice, narrowing a candidate set on probabilistic evidence. The cryptography is not reversed.
Dust-poisoning — sending tiny amounts to unrelated addresses to manufacture on-chain links, either to confuse analysis or to bait address-copying mistakes.
Hub — an address connected to a very large number of others, typically an exchange hot wallet. Merges unrelated clusters if not filtered.
Instant exchange — a swap service requiring no account, used to break the on-chain trail between two assets.
Two-hop link — a connection through two intermediate addresses. Present in the output because it can matter in combination, weighted low because on its own it rarely does.
Where to go next
Open Spectra — the platform itself
Training and certification — three-stage progression assessed on real casework
Intelligence On Chain — who builds it
Last updated





