Skip to content
CodesGuard

Autonomous testing

An autonomous red team that proves its findings.

A swarm of specialised agents maps your attack surface, reasons about what it found, chains small weaknesses into a real attack path and exploits it — streaming every decision and tool call live, then delivering a proof-first report you can replay.

Contact salesRead the docs
HighIDOR on invoice export — cross-tenant read Proven · 4 steps

Attack chain

  1. ReconRoute echoes a tenant id the client never sends

    httpx · jsluice

  2. DiscoveryExport endpoint absent from the OpenAPI spec

    ffuf

  3. AnalysisInvoice ids increment by one per tenant

    inference

  4. ExploitRequested a neighbouring id — got another tenant's PDF

    curl
Proof requestsession = tenant 8812

GET /api/invoices/44190/export HTTP/1.1

Authorization: Bearer eyJ…tenant_id=8812

Proof response200 OK · application/pdf

HTTP/1.1 200 OK

Bill to: Northwind Trading Ltd (tenant 9014)

Amount: 18,420.00 EUR

The question it answers

“What would a real attacker actually be able to do here?”

Capabilities

What AI Pentest does

A swarm of agents that reasons, chains weaknesses and proves the exploit.

A real kill chain

Reconnaissance, content discovery, vulnerability analysis, exploitation and reporting — each phase closes on evidence, never on a timer.

Live reasoning stream

Watch the swarm think, pick tools, run them and change plans as results come in. The full record is preserved for replay.

A full offensive toolkit

Reconnaissance, fuzzing, injection, secret discovery and exploitation tooling — dozens of capabilities the agents choose from based on the target in front of them.

Proof-first findings

A proven finding carries the request / response pair, the chain of steps that led there and a reproduction command. Suspected issues are labelled as such.

Safety rails

Runs only against DNS-verified targets in an attested engagement. Tool policy bounds what agents may do; destructive actions are blocked; one proof, then stop.

Profiles for every plan

Recon for surface mapping, Standard for full testing, Deep for extended exploitation with more delegations and revisits.

How it works

A real kill chain, phase by phase

01

Pick the target

Choose an engagement and a verified target. Attach a captured login if you want the swarm to test behind authentication.

02

Launch

There's no depth selector — the agents adapt their approach as they go and hand off between phases only on evidence.

03

Follow the stream

The run view shows reasoning, tool calls and findings as they're proven, phase by phase.

04

Read the report

Proven findings with evidence, attack chains, closing evidence per phase and a downloadable report.

The kill chain

Five phases, each closing on evidence

  1. 1

    Reconnaissance

    Surface mapping, DNS and certificate sweeps, technology fingerprinting.

  2. 2

    Content discovery

    Endpoints, parameters and JavaScript analysis; what the spec doesn't mention.

  3. 3

    Vulnerability analysis

    Auth, injection and business-logic probes against every discovered parameter.

  4. 4

    Exploitation

    Chained proof with evidence — one record touched, logged and attached.

  5. 5

    Reporting

    Verified findings, closing evidence per phase and a downloadable report.

DAST vs AI Pentest

Two ways to test a running system

They think differently — and they're complementary. Run DAST for coverage, AI Pentest to learn how far an attacker could actually get.

ApproachAutomated scanner — broad, systematic checks.Autonomous agents — reason, chain and exploit.
StrengthBreadth and repeatability.Depth: multi-step attacks a scanner won't find.
OutputVulnerabilities with evidence.Proven, chained exploits with a reasoning trail.
Best forRegular coverage of the attack surface.Understanding real-world exploitability.

Profiles

Tuned to the plan and the job

Profiles set how aggressive a run may be — the tool tier, whether exploitation is allowed, and how many times the swarm may delegate and revisit.

Recon

Free trial

Surface mapping only. Recon tooling, no exploitation, a handful of delegations.

Standard

By arrangement

Full kill chain with the standard tool tier and up to 15 delegations per run.

Deep

Enterprise

Extended exploitation: deep tool tier, exploit suggestions, up to 50 delegations and 20 stage revisits.

Available on

AI Pentest is included on the plans below.

Free trialStarterProEnterpriseTalk to sales →

FAQ

AI Pentest, in brief

Full walkthroughs live in the AI Pentest documentation.

How is this different from DAST?

DAST runs broad, systematic checks and is great for regular coverage. AI Pentest reasons: it combines small issues into a real attack, pivots from one foothold to the next and proves impact. Run both.

Is it safe to point at production?

Every run sends real attack traffic, so CodesGuard requires an authorization attestation and DNS ownership verification first. Tool policy and scope bound the run, destructive actions are blocked, and the agent stops after one proof.

How long does a run take?

As long as the work takes. Phases close when their evidence is complete, and you can cancel at any time. Plans set a ceiling on run duration and monthly run minutes.

Next module

SAST — Static analysis

Ship software you can prove is secure.

See the platform on your own code, or talk to us about a dedicated Enterprise stack.

Contact salesExplore the platform