SEC-2026 / CONTRACT SECURITY OPERATIONS
DOC. SEC-2026 / 11 TRAINED AGENTS / REV 2.6

Ship secure
contracts.

We are a security operations unit for your code. Paste a contract and eleven trained agents sweep every line — real line numbers, severity-tagged findings, and a fixed file you download when the sweep is done.

securithm — agent run
>
ReentrancyAgentAuthAgentLifecycleAgentContextAgentReturnValueAgentGasAgentTemporalAgentEntropyAgentPrivilegeAgentArithAgentGovernanceAgent
0.4s
TO VERDICT
0
TRAINED AGENTS
0
CHAINS WATCHED
24/7
MONITORING

[ WHAT WE ARE ]

A security operations unit
for your code.

DETECT01

11 agents trained on single vulnerability classes, cross-checked by static analysis. No black box — every finding names its agent and its line.

REPAIR02

Deterministic patches, not suggestions. Guards inserted, auth hardened, loops bounded — then handed back as a compilable file.

WATCH03

Deployed contracts monitored on six chains. Exploit attempts, governance anomalies and oracle drift alerted in under a second.

PROVE04

Proof-of-solvency attestations, signed reports and exportable audits — the paperwork your users and regulators ask for.

One protocol, three instruments

Everything between
your code and the exploit.

SCAN

Scan before you ship.

Paste a file, point at a repo, or drop a deployed address. A family of trained security agents reads your contract line-by-line and returns severity-tagged findings in seconds.

Run a free scan

FIX

Fixes, not lectures.

Every auto-fixable finding ships with a deterministic patch. Download the fully repaired .sol file, or a per-category unified diff you can apply in your repo.

See a fixed contract

MONITOR

Watch what's live.

Deployed contracts are watched around the clock across six chains. Exploit attempts, governance anomalies and oracle drift trigger alerts in under a second.

Open monitoring

The roster

Meet the agents
on your side.

Each agent is trained on one vulnerability class and reports with its own signature. Auto-fix agents patch the line they flagged; review agents leave annotated guidance for your team.

auto-fix review .sol / .patch
ReentrancyAgent
SENTINEL-01
auto-fix
AuthAgent
SENTINEL-02
auto-fix
LifecycleAgent
SENTINEL-03
auto-fix
ContextAgent
SENTINEL-04
auto-fix
ReturnValueAgent
SENTINEL-05
auto-fix
GasAgent
SENTINEL-06
auto-fix
TemporalAgent
SENTINEL-07
review
EntropyAgent
SENTINEL-08
review
PrivilegeAgent
SENTINEL-09
review
ArithAgent
SENTINEL-10
auto-fix
GovernanceAgent
SENTINEL-11
review
01

Scan

Paste your contract or point at a repo. Eleven agents sweep every line and grade the file A–F.

02

Fix

Auto-fix agents rewrite the flagged lines: guards inserted, auth hardened, loops bounded.

03

Download

Take the repaired .sol, the full unified patch, or a per-category diff — straight from the scan page.

[ TERMINAL ACCESS ]

Your pipeline,
hardened.

The CLI runs the same eleven agents from your terminal. Five free scans, no account. After that, one API key — generated at checkout — unlocks unlimited runs and dashboard sync.

terminal — securithm
# install
$ npm install -g securithm
# scan + fix
$ securithm scan Vault.sol --fix
# unlimited + sync
$ securithm login
paste key from dashboard — done.

Common questions

Asked, answered.

Anything written in Solidity, Vyper or Rust/Anchor — a pasted file, a GitHub repo, or a live deployed address. Trained rule agents plus static analysis cross-check every finding before it reaches you.

No. The first 5 scans are free and require no signup. Creating an account keeps your scan history private to your login, unlocks fix downloads, CI/CD integration and team seats.

The scanner asks you to pick a plan. Complete checkout — payment handled by Razorpay — and an API key is generated on the spot. Paste it into the CLI or use it on the site and scanning continues, unlimited.

Yes. After a scan completes you get the fully auto-fixed .sol file, a unified diff with every applied fix, and a per-category patch for each individual finding.

Yes. The GitHub Action scans every push and pull request, posts inline comments on findings, and fails the build above your configured severity threshold.

Run your first
scan free.