Skip to content
AgentsIndex

AgentRidge

Verified owner

Desktop agent that runs authorized penetration tests locally and exports a client-ready PDF report.

AgentRidge is a desktop application from Azuro Software that runs authorized penetration tests on e-commerce sites and web platforms, using a principal agent that delegates to specialized sub-agents. Every confirmed finding arrives with commands, evidence and a PDF report generated on your own machine.

Best for
Agent Ridge is best for e-commerce and web platform owners who need to find and prioritize exploitable vulnerabilities before launch or after a major update.
Look elsewhere if
  • you want to use it for real without paying
  • you want any plan without talking to sales
  • you need to cancel any time on every plan
  • you need a public API

What we checked

Pages last read
Facts checked
20 facts

From 17 sources, read 20 October 2026

AI answers
Too few read

58 buyer-question answers read, November 2026

AgentRidge in our research

All research

Study

AI Tool Pricing Index, November 2026

Pricing

One-time fee

3 of the 361 tools we checked charge a one-time fee.

Edition price, checked 20 October 2026.

Pricing pages of the AI tools, November 2026 editionas of n = 361 tools checked

A local desktop agent that runs the whole pentest

AgentRidge is a desktop application for macOS, Windows and Linux, not a cloud scanner. You open a New Mission, fill in a client name, a target IP or URL, optional VHost, an optional network CIDR and any operator context such as credentials to try, then pick engagement cases and a mode. The launch button stays disabled until the brief is valid, and the mission will not start until you confirm that testing is authorized for that target.

From there it works on its own. An orchestrator called Agent Ridge Principal plans the engagement and delegates to Ridge Sub-Agents for recon and enumeration, exploitation and verification, and parsing of raw tool output. The vendor is explicit that this is not a suggestion engine: "It's autonomous. The agent runs the full path end to end". If a sub-agent stalls or its local reasoning step fails, the Principal continues the work itself instead of ending the run. Output is a live attack-surface map, a vulnerability register, a command-by-command Activity feed and a PDF report.

Auditor and Attacker are two different products in one app

The mode you choose sets how far the agent goes. Auditor is described as proportionate and compliance-oriented, stopping at the minimum proof needed to raise a finding and avoiding noisy brute force. Attacker is maximum-depth adversary simulation that chains findings toward account takeover, privilege escalation and lateral movement.

The split matters commercially, because "Attacker mode is a Pro feature" and the free trial runs Auditor only. Both modes are said to avoid irreversible harm: no data wiping, no ransomware-style behavior, and no denial-of-service that risks production unless scope explicitly allows it.

The docs also devote a page to removing the refusal behavior that stops general assistants from doing offensive work, through operator-role framing, instruction precedence and authorization context. The specific implementation is proprietary, and the only control described is the written mandate the app forces you to confirm.

Three prices, two of them services rather than software

Pricing is flat and published: "500 € per audit" for an assessment the Azuro Software team runs for you, 500 € per service for remediation afterwards, and 2 000 € per year for the software licence you operate yourself. There is no monthly option and no billing toggle on the page.

Only the licence is self-serve. Both services start by preparing a request on the pricing page and sending it from your own email client, or writing to contact@join-ridge.com. The licence is bought in the browser, with Stripe handling payment and renewal governed by whatever the checkout link states; no refund or cancellation window is published. Model usage is included: the terms say the licence covers the included models without API consumption billing from Azuro Software.

The free trial is where the vendor's own pages disagree. The site FAQ promises "3 auditor-mode audits and 10 AI chats per monthly period", the documentation says only a limited number of audits per month, and the terms state the trial may include one Auditor audit with a knowledge pack of roughly 50%.

Local-first is the security story, and the only one

The reasoning core, called the Ridge inference engine, runs on your workstation. Targets, findings, evidence and PDF reports stay in a per-user directory: ~/Library/Application Support/AgentRidge on macOS, %APPDATA%\AgentRidge on Windows, ~/.config/AgentRidge on Linux. Sign-in and licence verification are the only cloud calls, which is a real answer for buyers who cannot send engagement data to a third party.

That also sets the cost of entry. The installer is small but the runtime downloads on first launch, about 25 GB free disk and 16 GB of RAM recommended, and on macOS setup asks for both Accessibility and Full Disk Access so the agent can drive tools in the background.

Nothing else is published. No SOC 2 report, ISO 27001 certificate or DPA appears on the site; the privacy policy lists HTTPS, access control and environment separation as its security measures. Account sign-in is email and password through the auth provider, with no SSO, and the terms offer availability "en l'état" with no uptime commitment.

Azuro Software, Paris, and an acquisition banner

The publisher is Azuro Software, 6 rue d'Armaille, 75017 Paris, with VAT number FR 35103328126 named in the legal notice. The stack behind the account is Supabase for authentication and database, Stripe for payment and Cloudflare for hosting. Google Analytics 4 and Google Ads tracking are active by default and refusable through the cookie banner, which is unusual for an EU site.

A banner across every page states that Agent Ridge was acquired by Altez Group, described as a 200M euro plus European leader. No press release, date or further detail about that acquisition appears anywhere on the site or in the documentation.

The public GitHub repository distributes desktop builds, CLI archives and checksums, and says plainly that it publishes releases and teaser documentation only. The CLI is a native binary with no source code, so this is closed software despite the GitHub presence.

Who should not buy it

Anyone wanting an API or an agent-callable surface. The only programmatic entry point is the ridge command on macOS and Linux, documented on the download page with a handful of subcommands. There is no HTTP API, no MCP server and no webhook, so results cannot flow into a ticketing system or a SIEM without manual work.

Teams, too. The licence is a single annual price with no seat model, no SSO and no account-level record of who ran what: the Activity feed audits a mission, not a user. A thin or managed laptop fleet is a poor fit for a 25 GB local runtime that wants Full Disk Access.

And anyone without a written mandate. Both the terms and the docs make authorization the buyer's responsibility, the app enforces a confirmation before launch, and the findings themselves can contain false positives that the vendor expects a human to validate before acting.

AgentRidge price and terms

Terms

Free to start
PartlyLimited free use
Card to start
YesNot required
Sign up without sales
PartlySome plans
The terms in full

Free account opens an Auditor-only trial: the site states 3 auditor-mode audits and 10 AI chats per monthly period, with quotas resetting monthly. Attacker mode, LAN discovery, detailed audits and the full runtime library are Pro only. No card is taken to start. The terms describe a narrower trial of one Auditor audit and a partial knowledge pack.

For AI agents

AgentRidge's pricing page

Pricing page, captured

join-ridge.com/en/pricing, checked

What we checked on AgentRidge

YesPartlyNo

What it costs to startPartly

Free to use

A free account opens an Auditor-only trial with monthly audit and chat quotas; Attacker mode, LAN discovery and the full runtime need Pro.

AgentRidge has a free trial and a Pro license. You start on the trial when you create your account during setup, and can upgrade to Pro at any time.

docs.join-ridge.com/reference/plansChecked

What it costs to startPartly

Self-serve signup

The €2,000 (about $2,198) license is bought in-app through browser checkout, but the €500 (about $550) audit and €500 (about $550) remediation start with an email request to the team.

Checkout is handled in your browser through the AgentRidge account system; once your subscription is active, the Pro features unlock automatically

docs.join-ridge.com/reference/plansChecked

What it costs to startNo

Card to start

Creating the account is what starts the trial, and payment only appears if you choose to upgrade at the Plan step or later.

Create a free account or sign in. Creating an account starts your free trial.

docs.join-ridge.com/guide/onboardingChecked

What it costs to startPartly

Cancel any time

Managed audit services are one-off, while the independent software offer is an annual Pro license without published cancellation terms.

Pro €2,000 / year (about $2,198)

docs.join-ridge.com/reference/plansChecked

What it works withNo

Public API

Automation is limited to the ridge command-line binary for macOS and Linux; no HTTP or REST API is documented anywhere in the product docs.

Install AgentRidge in your terminal with the ridge command.

join-ridge.com/en/downloadChecked

What it works withNo

MCP server

No MCP server or connector appears in the documentation, whose only cloud functions are sign-in and license verification.

Sign-in and license verification are the only functions that talk to the cloud.

docs.join-ridge.com/reference/plansChecked

What it does on its ownYes

Acts on its own

The agent plans, runs tools and pushes each lead to a conclusion without step-by-step approval, and the docs contrast this with assistants that only suggest commands.

It's autonomous. The agent runs the full path end to end and pushes each lead until the engagement's definition of done is met.

docs.join-ridge.com/reference/faqChecked

What it does on its ownYes

Limits on what it may do

Scope is set before launch through targets, engagement cases and mode, and the app blocks the mission until written authorization is confirmed.

Before a mission can start you must confirm that penetration testing is authorized for this target.

docs.join-ridge.com/guide/new-missionChecked

What it does on its ownYes

No code needed

Setup and missions run through a guided desktop interface, and the vendor states no security expertise is needed to read the findings.

The first time you open AgentRidge, a guided setup takes you from a fresh install to a ready-to-use Mission Control.

docs.join-ridge.com/guide/onboardingChecked

What happens to your dataPartly

No training on your data

Mission artifacts never reach the vendor unless you send them to support, so there is nothing to train on, but no explicit no-training clause is published.

Les artefacts de mission (cibles, findings, rapports PDF, journaux d'outils) sont générés et restent sur le poste local de l'utilisateur

join-ridge.com/confidentialiteChecked

What happens to your dataPartly

Retention you control

Local findings and reports are yours to delete with the data directory, while account data follows fixed policy periods and a 30-day erasure on request.

Compte inactif / suppression : effacement ou anonymisation sous 30 jours après demande, sauf obligations légales

join-ridge.com/confidentialiteChecked

What happens to your dataPartly

Data stays in your region

Engagement data sits in a per-user directory on your own machine, but the account, license and payment services offer no region choice.

AgentRidge stores its local runtime and your engagement data in a per-user application data directory.

docs.join-ridge.com/reference/data-loca…Checked

What happens to your dataYes

Self-hosted

There is no hosted alternative: the reasoning engine, the tooling and the report generation all execute on the workstation you install it on.

The reasoning core, the Ridge inference engine, runs locally rather than sending your engagement to a remote service.

docs.join-ridge.com/how-it-works/local-…Checked

What happens to your dataYes

Take your data out

Findings, evidence and generated PDF reports live in a documented per-user directory you can copy or back up, and reports export from the Reports view.

When a mission finishes, the generated PDF appears in the Reports view, where you can open it.

docs.join-ridge.com/guide/reportsChecked

Whether a company can buy itNo

Security audits

The privacy policy lists only general measures such as HTTPS and access control; no SOC 2, ISO 27001 or other audit is named on any page.

Mesures raisonnables : HTTPS, contrôle d'accès, secrets hors code source, séparation des environnements.

join-ridge.com/confidentialiteChecked

Whether a company can buy itNo

Single sign-on

Account data is an email address and a password held by the auth provider; no SAML, OIDC or identity-provider sign-in is documented.

Adresse e-mail, mot de passe (hashé côté fournisseur d'auth), nom (optionnel)

join-ridge.com/confidentialiteChecked

Whether a company can buy itPartly

Audit log

The Activity feed records every tool and command the agent runs during a mission, but it is a per-mission local trail, not an account-level record of who did what.

Every tool it announces and every action it takes is recorded here, so an engagement is fully auditable while it happens

docs.join-ridge.com/guide/activityChecked

Whether a company can buy itNo

Uptime promise

The terms promise reasonable availability on an as-is basis and commit to no uptime figure or service credit.

Le site et les services cloud associés sont fournis « en l'état » avec une disponibilité raisonnable.

join-ridge.com/conditionsChecked

Models it can useNo

Its own model

The reasoning engine ships under Ridge branding but the terms call the Principal a commercial model and the docs call the whole product white-label.

AgentRidge is a white-label product; inside the app the components are named Agent Ridge Principal, Ridge Sub-Agents, and the Ridge inference engine.

docs.join-ridge.com/reference/faqChecked

Models it can useNo

You pick the model

Setup provisions the reasoning engine itself; the documented onboarding steps and the plan table contain no model selection of any kind.

AgentRidge provisions the reasoning engine that powers the Principal and its sub-agents.

docs.join-ridge.com/guide/onboardingChecked

Questions about AgentRidge

How much does AgentRidge cost?
AgentRidge costs €500 (about $550) per audit. From join-ridge.com/en/pricing, checked 18 November 2026.
Does AgentRidge have an API or an MCP server?
AgentRidge has no API. From join-ridge.com/en/download, checked 20 October 2026. AgentRidge has no MCP server. From docs.join-ridge.com/reference/plans, checked 20 October 2026.
Is AgentRidge open source, and can you host it yourself?
You can host AgentRidge yourself. From docs.join-ridge.com/how-it-works/local-first, checked 20 October 2026.
What have we checked on AgentRidge?
20 facts, under these headings: What it costs to start, what it works with, what it does on its own, what happens to your data, whether a company can buy it, models it can use.
How do we check AgentRidge?
We hold 20 facts on AgentRidge, quoted from 17 pages. Facts last checked 20 October 2026. Price last checked 18 November 2026. Last change found 14 November 2026.

How we check every listing

AgentRidge price history

Plans for AgentRidge started at €500 (about $550) per audit on both of our reads, from 17 October 2026 to 18 November 2026.

PricePlanFirst seenLast seenReadingsSource
€500 (about $550) per audit17 October 202618 November 20262Pricing page, read 17 October 2026

From our research, read twice since 17 October 2026. 2 more of our checks could not pick this price out of the page and are left out. A date is when we read the page, not when the price changed.

Similar to AgentRidge

Claimed by the AgentRidge team.

We check its facts against the AgentRidge site.Facts checked 18 November 2026

Report an error if a fact is wrong.