DAEV / Code security auditor

Find the real cause of a code vulnerability and prove the smallest supported change that cuts the risky path.

A concrete example

A reachable checkout vulnerability

Input

A supported finding and the checkout repository inside the organisation's Athena deployment.

Action

DAEV traces how untrusted input could reach the vulnerable point and works out the smallest set of changes that closes every such path.

Result

The reviewer receives a proposed severing edit, a recomputable verdict, and an evidence graph.

What you receive

// 01

Supported path verdict

A precise account of the reachable paths the analysis established.

// 02

Severing edit

The smallest supported change that breaks those paths.

// 03

Evidence graph

Tamper-evident material the customer can retain and verify without DAEV running.

How it works

01
Load the supported target
Analyse the repository inside the customer's Athena deployment.
02
Trace the reachable paths
Follow how untrusted input could move through the code to where it could do harm.
03
Find the smallest fix
Work out the smallest supported set of changes that closes every established path.
04
Return the proof
Send the finding, edit, evidence graph, and verifier material into Athena for review.

Inside the mechanism

// 001

Reachability analysis

DAEV works out how untrusted input could reach the vulnerable point. The verdict comes from that analysis, not from a language model deciding whether the change is correct.

// 002

Smallest severing change

Where the analysis permits it, DAEV returns the smallest set of changes that closes every reachable path, with the evidence behind the proposal. The claim is limited to what the supported analysis represents.

// 003

Offline-verifiable evidence graph

The evidence graph packages the finding, the reachable path, the severing change, and verifier inputs so the customer can recompute the supported result without a live DAEV service.

What it proves

Bounded claim

For the supported analysis, the returned change closes every established reachable path the analysis represents.

What it does not prove

Permanent boundary

DAEV proves supported reachable paths, not that code is safe.

How it connects to Athena

Input

Repository reference, supported target, task identity, and correlation identifier.

Output

Finding, severing proposal, evidence graph, and verifier material.

Shared record

Athena keeps the proposed change, named decision, bounded execution, result, and evidence under the same record.

Continue the review

Follow runtime evidence and the visual analysis surface