> ## Documentation Index
> Fetch the complete documentation index at: https://docs.blockgence.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Investigation

> Trace funds on a canvas, save the case with its sealed evidence, and analyze a theft from the victim, the attacker or the transaction.

Investigation is the forensic tracing product. You start from an address, Blockgence follows the money forward and
backward and draws it as a graph, with the identity and risk of each address. Everything you see comes with its source
and method, and whatever could not be read is declared: a source outage is never shown as "no activity".

> [!NOTE]
>
> The app's interface is in Spanish. Screens, tabs and buttons are named as they appear, with their English meaning.

Open it from **app.blockgence.com → Investigación**. It has four sections:

> [card] Tracing canvas (Investigación)
>
> The fund graph of an address, with the timeline, the transactions table and the detail of each node.

> [card] Cases (Casos)
>
> Saved investigations, shared with your organization, with their sealed evidence, off-ramps and freeze request.

> [card] Incidents (Incidentes)
>
> Analysis of a theft or a hack from whatever you have: the victim, the drainer, the attacker, the protocol or the transaction.

> [card] Sources and health (Fuentes y salud)
>
> Which data sources are configured and what can be queried on each network.

![Investigation tracing canvas](https://mintcdn.com/blockgence/jIDqBW86tOU_Ini-/images/products/investigation/canvas.png?fit=max\&auto=format\&n=jIDqBW86tOU_Ini-\&q=85\&s=76055a9f8cfe2f122bb76effae6ca508)

The tracing canvas with demo data.

## Networks

Blockgence traces **Ethereum, BSC, Polygon, Arbitrum, Base, Optimism and Tron**. Other networks (Bitcoin, Solana and
others) only appear as the destination of a cross-chain transfer, with the address on the other side.

## Trace an address

**1. Pick the address and the network**

Paste the address in the search field (the <kbd>/</kbd> key takes you there) and pick the network. Only networks that
can be queried are listed.

**2. Adjust the parameters**

In **Parámetros** (parameters) you choose the forward hops (1 to 20), the branches per node (1 to 60), whether tokens are included,
whether the flow follows time order, a date window and the dust threshold in USD. Backward tracing covers 2 hops.

**3. Trace**

While tracing, a probe runs on the other networks (if the address has no activity on the chosen network, it tells you
where it does) and ENS names are resolved.

**4. Read the notice strip**

The tracing notices appear above the canvas: partial coverage (which source failed and where), activity on other
networks and, when it applies, the trail from a loss transaction or the stages of an attack. Notices push the graph
down: they never cover the data.

## Read the graph

* **Nodes.** Color is the category (sanctioned, mixer, scam, exchange, bridge, victim, contract, unknown and others),
  shape separates contracts from wallets, and size is value. The origin is highlighted.
* **Edges.** Color is the asset. Cross-chain transfers are blue (dashed when heuristic), swaps are teal and bridge refunds
  are orange. An edge with a mark (unverified, unknown decimals, refund) never loses its label.
* **Interaction.** A click opens the node detail and highlights the path from the origin. A double click expands one more
  hop. The **+N** hexagons reveal the counterparties the trace left out, 20 at a time.
* **Timeline and table.** Below the canvas, the timeline filters the graph by date and the table lists every flow, with
  its transactions, cross-chain transfers and swaps expandable. A click on an edge takes you to its row and back.

The **node detail** has six tabs:

| Tab | What it shows |
| - | - |
| Resumen (summary) | Confirmed risk and risk with heuristics, each with its reasons; findings (sanctions, bridge, the exchange where the trail goes cold); who funded it. |
| Flujos (flows) | Counterparties in the trace and hidden ones; at the origin, the exposure by category. |
| Identidad (identity) | The on-chain profile and the attribution: each label with its method, confidence and source. |
| Permisos (permissions) | EIP-7702 delegations, open approvals and third-party movements. All read-only. |
| Vector | How a wallet was compromised, by time window or by loss transactions. |
| Mensajes (messages) | The on-chain messages of the address. |

> [!NOTE]
>
> Risk separates what is **deterministic** (proven by on-chain data or a cryptographic id) from what is
> **probabilistic** (an inference with its confidence). Confirmed risk excludes the probabilistic part.

## Cases

A case saves the whole investigation: the starting address, the parameters, the graph with its expansions, the time
filter, the coverage and the verdicts.

**1. Create the case**

From the top bar. From then on it saves itself, a second and a half after each change.

**2. Open it again**

A case opens exactly as you left it, without querying the chain again.

**3. Share it**

From **Casos** (cases), with people in your organization, to view or to edit. When you open it, Blockgence tells you if its
addresses are already in other cases you can see.

**4. Export the evidence**

The evidence package is a sealed JSON with its sha256 and the Blockgence signature (an unsigned document says so),
and it can come with a PDF report. How to check it, in [Methodology and evidence](https://docs.blockgence.com/en/methodology/sealing.md).

A case also gives you the **off-ramps** (exchanges reached, downloadable as CSV) and the **freeze request**: the status
of each address with the stablecoin issuers that can freeze, their channels for authorities and the data ready to copy.

> [!WARNING]
>
> Blockgence does not send anything for you: it does not file requests with exchanges, issuers or authorities, it does
> not revoke permissions and it does not sign transactions. It leaves the data ready for you to file.

## Incidents

Paste what you have and say what it is. Blockgence shows the matching views:

| What you paste | What you see |
| - | - |
| A victim's address | The victim triage (how the funds were taken, where they went and what can be frozen), outflows in transactions the victim did not send, address poisoning and the compromise vector. |
| A drainer's address | The drainer's victims. |
| A scammer's address | The scam victims with evidence, kept apart from hints and unqualified counterparties. |
| An attacker's address | Failed attempts and attack stages. |
| A protocol's address | Whether it adopted SEAL Safe Harbor, at a given date. |
| A transaction hash | The loss transaction, the exploit analysis and the returns. |
| A case token | The post-mortem draft, with the SEAL structure. |

**Modo incidente** (incident mode) follows a playbook for the first hours: you paste the hash, the victim or the attacker's addresses,
set a credit cap and follow each step live, with what it charged. At the end it opens the case or the post-mortem.

![Incidents section](https://mintcdn.com/blockgence/jIDqBW86tOU_Ini-/images/products/investigation/incidents.png?fit=max\&auto=format\&n=jIDqBW86tOU_Ini-\&q=85\&s=b5badd54f35e86d13c1b3166e0194b8a)

Incidents: an address analyzed as a victim, with demo data.

## Sources and health

One card per data source: whether it is configured, its status, its license and its last error. Below, a matrix shows
what can be queried on each network. Use it to understand a partial-coverage notice before drawing a conclusion.

## Through the API

Everything in Investigation is in the API: tracing (`POST /v1/trace`), the address profile, funding, counterparties,
cross-chain transfers, swaps, cases (`/v1/cases`) and evidence. Each query can be tied to a case with the `X-Case-Id`
header.

```bash Trace theme={null}
curl -X POST https://app.blockgence.com/v1/trace \
  -H "X-API-Key: $BKG_KEY" -H "Content-Type: application/json" \
  -d '{"address": "0x00000000000000000000000000000000000a11ce", "chain": "ethereum", "hops": 3, "fanout": 6}'
```

The response is a graph (`nodes` and `edges`) with the label, category and explained risk of each node, and the
exchanges reached in `offramps`. Whatever could not be read travels in the same response. Details in the
[API reference](https://docs.blockgence.com/en/api/reference.md).

## What it does not cover

* **The graph is a sample**, bounded by hops, branches, dust and row caps. It is not the full history, and what is left
  out is declared.
* **Shared infrastructure** (routers, pools, vaults) cuts the trail on purpose: what comes out of it belongs to other
  users.
* A bridge that is not among the reconstructed ones is named but not crossed. A transfer to a network that is not
  queried is declared.
* Mixer **demixing** covers only Tornado Cash on Ethereum, and it gives leads, not proof.
* **NFTs** are not traced (they show up in approvals and in the compromise vector), and tokens without a published price
  are filtered out of the trace as spam, declared.
* Attribution comes from public sources, verified registries and your organization's labels. It is not a proprietary
  entity database at scale.
* Exploit heuristics do not cover business-logic bugs, external oracles, governance or compromised keys. Most heuristics
  declare that their error rate is not measured.
* **Monero** is not traced: the network does not publish senders, recipients or amounts. Only screening and the payment
  proof are available.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.