Pentest your code on every push.
Elliot runs Elliot-1, our offensive security model, against your repository and reports only the vulnerabilities it can trace from input to impact. Each one comes with the request that triggers it and the fix.
Traced from input to exploit
Each finding follows the attacker’s input through your code, so your agents know precisely what to fix.
A PoC for every finding
Each finding is tested in a sandbox environment before it hits your board. You only have to fix what’s really broken.
Vulnerable dependencies your code reaches
Elliot checks your npm and Python lockfiles against known advisories, and reports a package only when a path runs from an entry point to the vulnerable call.
Hand findings to your coding agent
Connect Claude Code, Cursor or VS Code to Elliot’s MCP server. Your agent reads the trace and the PoC, patches the code and moves the finding along your board.
We tested this and it immediately found a huge vulnerability for us. Can’t recommend it more.

What happens in a scan.
Elliot works through your repository in five passes. Findings stream onto your board while the scan is still running.
- 01
Index
Maps routes, handlers, callers and deploy config at the head of your branch.
- 02
Threat model
Works out what is worth stealing, where the trust boundaries are and how tenants are kept apart.
- 03
Hunt
Attacks every entry point from several angles, and digs again wherever it finds something.
- 04
Ground
Checks each step of every trace against your real code and merges reports that share a root cause.
- 05
Prove reach
Decides who can reach the bug and what they gain. If the answer is nothing, it stays off your board.
From push to fixed.
Scans on every push
Run daily, weekly or on push. Incremental scans read only the files that changed since the last one.
A board your team already knows
Triage, to do, in progress, in review, fixed. Assign a finding and move it along like any other issue.
Fixes get checked
Mark a finding fixed and the next scan looks for it again. If the bug is still there, it goes back to triage.
Exploits run in a sandbox
Elliot connects through a GitHub App that reads code and metadata. Proofs of concept run against your code in an isolated sandbox, away from your servers. Pull request checks are optional and add permission to post results.
Start with a free scan.
Connect GitHub, pick a repository and run a full scan for free. You see up to five findings in full, with the trace, the request and the fix. Choose a plan and the same scan carries on past them.
Free
$0
- 1 repository
- One full scan, up to five findings
- Each finding in full, with trace, request and fix
- Daily supply chain analysis
- Upgrade and the scan carries on
- Model: Elliot-1
Pro
$149per month
- 2 repositories
- 200 scans a month
- Then $1 per extra scan
- Up to 10 contributors
- Full findings with trace, request and fix
- Daily supply chain analysis
- Model: Elliot-1
Team
$499per month
- 10 repositories
- 1,000 scans a month
- Then $1 per extra scan
- Up to 25 contributors
- Full findings with trace, request and fix
- Daily supply chain analysis
- Model: Elliot-1
Enterprise
Custom
- Unlimited repositories
- Scan volume to match
- SAML single sign-on
- Audit logging
- Dedicated support channels
- Daily supply chain analysis
- Model: Elliot-1
Questions.
Does Elliot run my code?
Only inside a sandbox. To test a finding, Elliot writes a proof of concept and runs it against your code in an isolated sandbox. Your servers and deployments stay untouched.
What access does it need?
Read access to contents and metadata on the repositories you choose during setup. If you turn on pull request checks, Elliot also needs permission to post check results on pull requests. You can add or remove repositories from GitHub at any time.
How is this different from a static analysis tool?
Pattern-based tools flag code that looks dangerous. Elliot follows input from an entry point to the line that breaks, checks every step against your code and works out who can reach it before a finding lands on your board.
Do I still need a pentest?
Keep the annual pentest if a customer or auditor asks for one. Elliot covers the rest of the year: it scans on every push, and Pro is $149 a month.
What is Elliot-1?
Our model for offensive security work, adapted from DeepSeek V4.1 Flash. On our internal benchmark it passed 2 percentage points more tasks than its base model at 13.2% lower cost per task. Every plan runs it.
What if Elliot gets one wrong?
Mark it as a false positive and it stays closed, even when a later scan finds the same thing again.
What counts as a scan?
Each completed scan that reviews code. Past your plan’s scans for the month, each one is $1, added to your next invoice.
See what Elliot finds in your code.
Connect GitHub, choose a repository and Elliot runs its first full scan for free.
Run a free scan