> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tori.finance/llms.txt
> Use this file to discover all available pages before exploring further.
# Security Overview
> How Tori Finance approaches security
Tori runs layered security: independent audits before deployment, continuous monitoring in
production, third-party attestation of reserves, and privileged keys held behind multisigs and a
24-hour timelock. No single layer is trusted to hold on its own, and each one can be checked from
outside. This page describes the whole stack; the pages it links to carry the detail.
## The security stack
| Layer | Provider | Detail |
| - | - | - |
| Smart contract audits | Sherlock and Nethermind | [Audit reports](/security/audits) |
| Threat monitoring | Hypernative | AI-driven surveillance of protocol activity, around the clock |
| Proof of Reserves | Accountable | Live attestations at [tori.accountable.capital](https://tori.accountable.capital) |
| Key custody | Safes, MPC vaults, timelock | [Role matrix](/security/roles) |
## Smart contract audits
Sherlock and Nethermind audited the protocol independently. Sherlock has historically secured over **\$50 billion**
in assets across Web3; Nethermind is a blockchain research and security firm. The audits covered
the core protocol contracts, strUSD staking and unstaking, access control, and third-party
integrations.
Both reports are public, with dates and PDFs on the [audits page](/security/audits), alongside a
bug bounty that pays up to \$1,000,000 for a critical finding.
## Monitoring
Hypernative watches protocol activity continuously and alerts the team the moment something looks
wrong: an unusual or outsized transaction, a known attacker address, an admin operation, a
parameter change, an external dependency misbehaving. Its models score risk on an ongoing basis,
and the first response to a detected threat is automated.
## Proof of reserves
Accountable attests reserves and financials independently and continuously. The attestations cover
total assets under management, the liability coverage ratio, and reserve fund status, and they are
published at [tori.accountable.capital](https://tori.accountable.capital), where anyone can read
them at any time. The figures are third-party attested rather than self-reported, and they update
continuously instead of on a monthly or quarterly cycle. The app mirrors the live backing data at
[app.tori.finance/transparency](https://app.tori.finance/transparency).
## Custody
On-chain, the contracts are audited, critical operations take a multisig quorum, and anything that
widens the protocol's surface waits out a [24-hour timelock](/security/roles). Every function sits
behind a role scoped to a single job.
Off-chain reserves sit with vetted institutional custodians and partners, in segregated accounts,
diversified across jurisdictions. [Backing and reserves](/solution/backing) covers the asset side
in detail.
Protocol reserves are not commingled with operational funds. All assets are held with vetted
custodians, institutional partners, or through regulated instruments.
## Who can touch what
Every privileged role on the mainnet contracts is documented in the [role matrix](/security/roles),
down to the function level: what each role can do, who holds it, and how the key is custodied. The
short version: no role can do more than the one job it is named for, upgrades and new custody
destinations wait 24 hours in public before they can execute, and every role holder is readable
on-chain.
## Operational security
Team access requires multi-factor authentication, signing keys live in hardware security modules,
each person carries the minimum access their role needs, and access rights are reviewed on a
schedule.
When an incident is detected, the process runs: triage and severity classification, containment,
remediation, communication to users, and a post-mortem. Containment does not wait on a multisig quorum: the
Gatekeeper role can shut off mint and redeem in a single transaction.
## Reporting a vulnerability
Report security issues to [security@tori.finance](mailto:security@tori.finance). We aim to acknowledge
reports within 24 hours. Rewards, scope, and disclosure guidelines are on the
[audits page](/security/audits#bug-bounty).
No system can guarantee perfect security. Audits are point-in-time assessments, and new attack
vectors emerge. That is why the layers above overlap. See [Risk disclosures](/resources/risks) for
the complete picture.
## Next steps
Both reports, with dates, PDFs, and the bounty table
Every privileged key, down to the function level
What stands behind trUSD
What can go wrong, stated plainly
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.