Book a demo

Build at AI speed.
Stay in control.

FlowDex turns your entire codebase — code, infrastructure and configuration — into living documentation, end-to-end flows, change impact and security insight. For your developers, your product owners and the AI agents that write code with them.

Book a demo See it in action
  • SaaS or fully self-hosted
  • Bring your own AI model
  • MCP for Copilot, Claude & Cursor
Live preview · demo data
Your applications and how they connect HTTP HTTP BookingCreated email capture reserve Web appBooking site APIBooking API ServicePayments Azure Service Bustopic: bookings FunctionNotifications ExternalHotel partners Focus areas of Booking API What the application does, grouped by business capability Booking & checkoutCreate, change and cancel bookings18 flows Payments & refundsTake payments, refund cancellations11 flows Search & availabilityFind trips and check availability14 flows Customer accountsSign-up, profile and travellers12 flows Flows in Booking & checkout Book a tripPOST /bookingsOpen → Cancel a bookingDELETE /bookings/{id} Show my bookingsGET /bookings Change travel datesPUT /bookings/{id}/dates Retry a failed paymentEvent · PaymentFailed Applications Focus areas Flows

Works with the stack you already run

  • GitHub
  • Azure DevOps
  • GitLab
  • Bitbucket
  • VS Code
  • Visual Studio
  • JetBrains Rider
  • WebStorm
  • Cursor
  • Windsurf
  • Zed
  • GitHub Copilot
  • Copilot CLI
  • Claude Code
  • Claude Desktop
  • Codex CLI
  • Gemini CLI
  • Cline
  • Any MCP client
  • .NET / C#
  • TypeScript
  • Azure Service Bus
  • MassTransit
  • NServiceBus
  • Jira
  • SonarQube
  • Trivy
  • OpenAI
  • Azure OpenAI
  • Anthropic
  • Google Gemini
  • Ollama
300repositories documented in one enterprise landscape
4×feature output in pilot teams*
65%less engineering effort per feature*
€3.7Msaved yearly for a team of 80 engineers*

* Pilot results and model calculation: 80 engineers × 1.5 hours saved per week on documentation and research (€414,000) plus the 65% less effort on feature work seen in pilots (€3,318,900), at €75 per hour. Run your own numbers

The challenge teams face

AI accelerates development. It also breaks ownership.

AI assistants produce more code than ever, but understanding hasn't kept pace. Pull requests grow, context disappears and teams increasingly rely on assumptions instead of understanding.

“This PR is pretty big… but we're going to test it anyway.”Pull request review

“I think it works… but I don't know what else this relates to.”Refactoring

“Who actually built this code?”Incident review

“I actually don't dare to touch this part.”Legacy module

Does anyone fully understand this?

  • Build faster, understand less. Less self-written code means less ownership.

  • Features get duplicated or built incorrectly because nobody sees what already exists.

  • Pull requests are reviewed without full understanding of what they touch.

  • Changes go live with uncertainty about their impact.

  • Engineers don't fully trust the output of AI.

How do we keep control and trust in a world where AI writes code faster than we can understand it?

Speed without control = risk. The bottleneck has moved from building to understanding.

How FlowDex works

Take control of your codebase in five steps.

FlowDex makes complex software understandable, with full insight and control. The result is documentation that serves humans and AI alike — and stays current with every scan.

Documentation for humans and AI
  1. Step 1 · Index

    Indexes the entire codebase

    FlowDex scans source code, infrastructure and configuration across all your repositories — including private packages — and builds a deterministic graph of entrypoints, calls, messages and data access.

    .NET / C#TypeScriptInfrastructure as codePrivate packages
  2. Step 2 · Document

    Builds complete application documentation

    It maps how requests flow through the system, how components interact and how business processes work — described both functionally and technically, per flow, per focus area and per application.

    End-to-end flowsSequence diagramsArchitecture maps
  3. Step 3 · Enrich

    Adds smart context

    AI enriches the graph with summaries, focus areas, business rules, risks, dependencies and opportunities for reuse. Every claim links back to code evidence; what can't be proven is shown as a gap.

    Evidence-backedBusiness rulesConfidence score
  4. Step 4 · Detect change

    Shows the impact of every change

    New deployment or pull request? FlowDex immediately shows what changed, which flows are affected and what needs to be retested — integrated into your CI/CD pipeline.

    FlowDex DIFFRetest adviceQuality gate
  5. Step 5 · Deliver

    Makes everything immediately usable

    Developers, product owners and AI agents use the same knowledge through the FlowDex app, the API and the MCP server — including agent workflows that build features, fix issues and add tests.

    Web appAPIMCP serverAgent workflows

Product tour

See what FlowDex actually shows you.

A look inside FlowDex with a fictional codebase: Contoso Travel, a booking platform with 14 projects, a TypeScript frontend, .NET services and Azure Service Bus. Every screen below uses demo data — yours is generated from your own code.

flowdex · contoso/travel-platform · main
Demo data
Projects contoso/travel-platform Booking API
Flow insight cockpit
What does Booking API do?

Booking API lets travellers search trips, book and pay, and manage their bookings. Confirmations and payment capture run asynchronously over Azure Service Bus; reservations with airlines and hotels go through the Partner Gateway.

6Focus areas
71Flows explained
48API routes
15Async side effects
94%Confidence

Booking & checkout

Create, change and cancel bookings, from checkout to confirmation.

18 flows12 API routesasync

Payments & refunds

Take payments, handle provider webhooks and refund cancellations.

11 flows6 API routesasync

Search & availability

Find trips, hotels and flights and check live availability.

14 flows9 API routesread-only

Pricing & vouchers

Calculate prices, apply vouchers and seasonal surcharges.

9 flows5 API routes

Customer accounts

Sign-up, profile, fellow travellers and marketing consent.

12 flows10 API routes

Partner integrations

Sync inventory and reservations with airline and hotel suppliers.

7 flows3 jobs2 gaps
Booking & checkout Book a trip
Flow detail
Book a trip
Entrypoint · POST /api/bookings94% confidenceasync follow-up work7 tests1 gap
What this flow does

A traveller confirms a trip in the checkout. Booking API validates the travellers and the price quote, reserves seats or rooms at the partner and stores the booking as Pending. It then publishes BookingCreated, so payment capture and the confirmation email happen asynchronously.

Business rules and functional conditions 12
  • PreconditionAt least one adult traveller is required.View code
  • PreconditionThe price quote must be less than 15 minutes old.View code
  • Guard / deviationIf the partner cannot reserve, the booking is rejected with 409 Conflict.View code
  • ResultThe booking is stored with status Pending.View code
  • Side effectBookingCreated triggers payment capture and a confirmation email.View code
  • Known gapNo timeout is configured for the partner reservation call.View code
Flow diagram
Checkout Booking API Pricing Partners Database Service Bus 1POST /api/bookings 2validate quote 3reserve inventory 4save Booking (Pending) 5BookingCreated → Payments, Email 6201 Created
Input
  • quoteId Guid · required
  • travellers TravellerDto[] · required
  • voucherCode string · optional
Output
  • 201 BookingResponse
  • 400 validation errors
  • 409 not available
Database writes / reads
  • CreatesBooking, Traveller
  • ReadsPriceQuote
  • UpdatesInventoryHold
Test coverage for this flow
  • 7 tests cover 12 functions
  • 82%
Pull requests #1482
FlowDex DIFF · change insight
Apply vouchers before the seasonal surcharge
Openfeature/voucher-order → main+128 −41 · 4 filesHigh impact
7Flows affected
3Entrypoints changed
2Focus areas touched
2 / 14Projects to build & test
Affected flows ranked by impact
FlowImpactRetest
Book a tripPOST /api/bookingsHighRetest
Get a price quotePOST /api/quotesHighRetest
Apply a voucherPOST /api/bookings/{id}/voucherHighRetest
Refund a cancellationEvent · BookingCancelledMediumRetest
Change travel datesPUT /api/bookings/{id}/datesMediumRetest
Monthly revenue reportJob · nightly 02:00Low—
Sync partner pricesJob · every 15 minutesLow—
Blast radius from the changed method
PricingCalculatorCalculate() · changed Book a tripPOST /api/bookings Get a price quotePOST /api/quotes Apply a voucherPOST …/voucher Change travel datesPUT …/dates Refund a cancellationEvent · BookingCancelled +2 low-impact flows
Business rule changed. Refund a cancellation now refunds the discounted price instead of the list price. Check with Payments & refunds before merging.
Smallest safe test scope: build and test Booking.Api and Pricing.Engine — 38 tests instead of the full suite.
Security overviewThe most complex method scores 31, above the limit of 15.
Quality gate failedSecurity scan completed
Projects145 with CVEs · 8 with code findings
Dependency CVEs171 critical · 5 high
Code quality38Highest complexity 31 · limit 15
Security findings61 secret · 2 license · 3 misconfig
SBOM1,28486 direct · 1,198 transitive packages
Dependency CVEs 17 findings in 11 package versions
Critical 1High 5Medium 8Low 3
Critical
System.Text.Encodings.Web4.5.0CVE-2021-26701 · remote code execution · used by 3 flows in Partner integrations
nugetFix 4.5.1
High
Microsoft.Data.SqlClient5.1.1CVE-2024-0056 · security feature bypass · 4 projects
nugetFix 5.1.3
High
Newtonsoft.Json12.0.2CVE-2024-21907 · denial of service via deep nesting · 9 flows
nugetFix 13.0.1
High
System.Text.Json8.0.4CVE-2024-43485 · denial of service · 2 projects
nugetFix 8.0.5
Medium
axios1.5.1CVE-2023-45857 · XSRF token exposed to third-party hosts · booking-web
npmFix 1.6.0
Security findings 4 of 6
Azure Storage account keySecret
src/Partner.Gateway/appsettings.Development.json:14Rotate the key and move it to Key Vault.
itext7 8.0.2 · AGPL-3.0License
Copyleft license in a closed-source service.
Container runs as rootMisconfig
deploy/booking-api/DockerfileAdd a non-root USER instruction.
Public blob access enabledMisconfig
infra/storage.tfDisable anonymous access on the storage account.
Booking API Code quality
Quality gate · main
Code quality and test coverage
Quality gate failed

2 of 6 conditions failed on this branch.

  • No new critical CVEs
  • Max complexity ≤ 15 (31 found)
  • No secrets in changed code
  • Coverage on new code ≥ 80% (71%)
  • No new duplicated blocks
  • All entrypoints resolved
Functions in “Book a trip” complexity · coverage
FunctionComplexityCoverage
BookingsController.Create4100%
CreateBookingHandler.Handle991%
PricingCalculator.Calculate3164%
VoucherRule.Apply1288%
PartnerGateway.ReserveAsync758%
BookingRepository.AddAsync2100%

Complexity: green <10, amber 10–14, red ≥15. Coverage: green ≥90%, amber ≥75%, red below.

Duplicated logic 92% similar

Refunds re-implement the cancellation fee. Reuse one instead of maintaining both.

Obsolete code still in use. LegacyPaymentClient is marked [Obsolete] but still runs in 2 flows: Retry a failed payment, Refund a cancellation.

Security & compliance

Know which business flows a vulnerability actually touches.

Most scanners hand you a list of CVEs. FlowDex tells you where they matter: which projects, which packages and which customer journeys use them — so you fix what's risky first.

  • Dependency and container CVEs, grouped per package version with the version that fixes them
  • Secrets in code and configuration files
  • License compliance: copyleft and commercial license risks
  • Misconfigurations in Dockerfiles and infrastructure as code
  • A complete SBOM of direct and transitive packages, ready to export
  • A quality gate for your pipeline, per branch and per project

Newtonsoft.Json 12.0.2

High · 7.5

CVE-2024-21907 · denial of service via deeply nested JSON. Fixed in 13.0.1.

Projects

Booking.Api nuget
Partner.Gateway nuget
Reporting.Jobs nuget

Flows that use it

Import partner feed Job
Book a trip POST
Payment webhook POST
Revenue report Job

3 projects · 9 flows · 2 focus areas — fixed with one upgrade. Demo data.

Grouped per upgrade

CVEs are grouped per installed package version, because that's how you fix them: one upgrade clears every advisory for that version.

Linked to your flows

See which entrypoints, jobs and message handlers use a vulnerable package — and which focus areas carry the business risk.

Fixed by your agent, safely

With /flowdex_fix_issues your AI agent fixes one finding at a time in an isolated branch, with checks before and after and a pull request as the result. How it works

FlowDex MCP for AI agents

Context is the new code. Give your agent the map.

AI assistants are great at writing code and bad at knowing your system. The FlowDex MCP server gives GitHub Copilot, Claude, Cursor and any MCP-compatible agent the understanding your best engineer has: existing flows, reusable components, impact and architecture.

  • Reuse before rebuild

    The agent searches existing flows and components first, so it extends what you have instead of duplicating it.

  • Impact before edit

    Before touching shared code it checks which flows, entrypoints and projects are affected.

  • Smallest safe test scope

    After the change it gets the exact projects to build and the flows to retest.

  • Fewer tokens, better results

    Precise answers from the graph instead of reading dozens of files: faster, cheaper and more consistent output.

overviewfocus_areasflow_insightfind_elementsimpactaffectedquality_findingsquality_retest_advicequality_coverage_test_planflowdex_fix_issues_startflowdex_test_coverage_start
See the agent workflows
Agent chat contoso/travel-platform · demo
Add a €25 fee when a booking is cancelled within 14 days of departure.
flowdex · focus_areasRelevant areas: Booking & checkout, Payments & refunds
flowdex · find_elements "cancellation fee"Found CancellationPolicy.GetFee() — already applies a 48-hour fee
flowdex · impact CancellationPolicy.GetFee4 flows · 2 projects · duplicate logic in RefundPolicy.CalculateFee
flowdex · affectedBuild & test Booking.Api and Payments.Service (2 of 14 projects)
Agent · with FlowDex context

I'll extend the existing CancellationPolicy.GetFee() instead of adding a new service — it already handles the 48-hour rule, so the 14-day window belongs there.

Heads-up: RefundPolicy.CalculateFee duplicates this logic, so refunds would ignore the new fee. I'll route both through CancellationPolicy.

  • Changes: 2 files, 1 new test
  • Retest: Cancel a booking, Refund a cancellation, Change travel dates
Without FlowDex164k
With FlowDex MCP19k
Tokens used for this task · illustrative example

FlowDex agent workflows

From feature to tested pull request. Your AI agent, on rails.

FlowDex MCP doesn't just answer questions. It gives your AI agent fixed, repeatable workflows: understand the context, turn a feature into stories, write the tests first, build on what already exists, compare before and after, and open a pull request with the evidence attached. Your agent programs faster because it starts with the answer instead of the search.

  1. 1

    Understand

    The agent reads focus areas, flows and business rules from FlowDex instead of crawling dozens of files.

    overview · flow_insight
  2. 2

    Plan stories

    A feature becomes sharp stories and technical tasks, with the questions nobody asked and the flows each story reuses.

    focus_areas · impact
  3. 3

    Tests first

    Tests are written up front from the extracted business rules and the flows to retest. They fail — as they should.

    quality_retest_advice
  4. 4

    Build, reuse first

    A mandatory reuse check runs before the first edit. The agent extends existing code and checks impact before touching anything shared.

    find_elements · impact
  5. 5

    Compare

    The same checks run before and after the change: tests, coverage, complexity, findings and the quality gate, plus the flows affected.

    affected · quality_gate
  6. 6

    Pull request

    One branch, one pull request, with the reuse decision, test results, gate evidence and retest list attached.

    git · PR
Agent · FlowDex workflow run Demo data

> /flowdex_implement contoso/travel-platform "Split a payment between travellers"

  1. Read context overview · flow_insight14 projects · 71 flows · "Capture payment" and "Payment webhook" explained
  2. Plan 3 stories, 7 tasks focus_areas · impact2 open questions answered by the product owner · 4 flows affected
  3. Reuse preflight find_elements · elementExtend PaymentIntent and PaymentWebhookHandler — no new service needed
  4. Write tests first quality_retest_advice9 tests from the business rules · 9 failing, as expected
  5. Build the smallest slice impact · affected5 files in Payments.Service and Booking.Api
  6. Compare before and after quality_gate9/9 new and 212/212 existing tests pass · coverage 82% → 88% · gate passed
  7. Pull request #1491 openedWith reuse decision, test results and 4 flows to retest

> /flowdex_fix_issues contoso/travel-platform maxIssues=1

  1. Select finding quality_findingsNewtonsoft.Json 12.0.2 · CVE-2024-21907 · High
  2. Show what's at stake quality_retest_advice9 flows use this package, incl. "Book a trip" and "Import partner feed"
  3. Isolated worktree flowdex_fix_issues_startBranch flowdex/fix/cve-2024-21907 · one fix-pack
  4. Baseline checksBuild and 212 tests pass before any AI edit
  5. Scoped AI fixUpgrade to 13.0.1 and adapt 2 call sites in Partner.Gateway
  6. Same checks, compared quality_gate212/212 tests pass · CVE resolved · no new findings
  7. Pull request #1493 openedOne finding, one branch, one pull request, with evidence

> /flowdex_test_coverage contoso/travel-platform

  1. Pick the riskiest gap quality_coverage_filesPricingCalculator.cs · 64% covered · complexity 31 · used by 7 flows
  2. Plan the tests quality_coverage_test_plan11 scenarios: vouchers, seasonal surcharge, rounding, currency
  3. Isolated worktree and baseline flowdex_test_coverage_start38 existing tests pass · coverage 64%
  4. Write tests on public behavior file · entrypoints · impact11 new tests · no production code touched
  5. Compare coverage test_report49/49 pass · coverage 64% → 91%
  6. Backlog updated quality_coverage_filesPricingCalculator.cs left the coverage backlog
  7. Pull request #1494 openedTests only, ready for review

Built-in guardrails

AI writes the code. FlowDex decides when it's done.

Every workflow runs the same fixed steps, every time. The AI only does the scoped edit; FlowDex owns the selection, the checks and the gate.

  • Reuse before new codeEvery answer states the reuse decision: extend what exists, or new code with evidence.
  • One change, one branch, one PREach fix runs in its own isolated worktree. Nothing is batched.
  • Green baseline firstChecks must pass before any AI edit, and again after it.
  • Evidence before a pull requestNo PR without passing checks and FlowDex gate evidence.
  • AI doesn't grade its own workThresholds, check results and gate outcomes are decided by FlowDex, not the model.
  • No secrets to AICredentials and private environment values never reach the model.

Fits how you already work

From feature request to pull request, with context at every step.

Features become sharp stories

Start a feature in Azure DevOps, Jira or your own tool. FlowDex asks the questions nobody thought of and creates stories and technical tasks that reuse existing flows.

Boards / Contoso Travel / Features
Feature

Split a payment between travellers

Let the lead booker split the total so every traveller pays their own share.

?

FlowDex asksShould each traveller get their own refund when the booking is cancelled?

?

FlowDex asksIf one share fails, do we hold the booking or cancel it after 24 hours?

User story

As a lead booker, I can split the total between travellers

Reuses: Capture paymentAffects 4 flows
Task

Extend PaymentIntent with payer shares · Payments.Service

Task

Reuse PaymentWebhookHandler for partial captures

Every pull request gets an impact check

FlowDex runs in your CI/CD pipeline. Reviewers see changed flows, affected entrypoints, what to retest and whether the quality gate passes — before anything goes live.

Some checks were not successful2 successful · 1 needs attention · 1 failing
build · Booking.Api — successful in 2m 14s
tests · 38 passed — scope selected by FlowDex
FlowDex · change insight — 7 flows affected, 5 to retestDetails
FlowDex · quality gate — complexity 31 > 15Details
Rescans are incremental. FlowDex compares every flow's evidence and only re-synthesizes flows that changed, so AI costs stay low.

Any Git provider

Connect GitHub, Azure DevOps, GitLab or Bitbucket, including private package feeds, branches and monorepos.

Your own AI model

Use OpenAI, Azure OpenAI, Anthropic, Google Gemini, Ollama or a custom LLM — whatever your security team approved.

Enterprise ready

Single sign-on, tenants and roles, audit trail and analytics, and self-hosted containers for full control.

From insight to results

One source of truth for everyone who shapes software.

1

Developers

Deliver and evolve software faster and with more certainty. Reviews get easier, impact is clear sooner and the risk of regression drops.

Onboard in days, not months.

2

Product owners

Sharpen the path from idea to execution. Features are defined faster, built more effectively and with less miscommunication between product and engineering.

Understand the product without reading code.

3

AI agents

Get the context that is normally scattered across code, flows and documentation. Output becomes more relevant, consistent and reliable.

Fewer tokens, better results.

4

Business

Lower failure costs and more predictable delivery. Teams go live faster without sacrificing control or reliability.

Knowledge stays when people leave.

Impact on delivery

From weeks to days, with fewer engineers.

AI-assisted teams without shared context spend their time on analysis, research and rework. With FlowDex, that time goes into shipping.

AI-assisted engineers without FlowDex
6 developers
5–6 features
per quarter
65%less engineering effort
4×output
AI-assisted engineers with FlowDex
2 developers
20–25 features
per quarter

Based on a FlowDex pilot team working on an existing enterprise codebase. Results depend on codebase and team. Calculate your savings

Self-hosted or SaaS · bring your own model

Your code. Your control.

FlowDex works within your existing security and governance frameworks — not around them.

Flexible deployment

Run FlowDex as SaaS, in your private cloud or fully self-hosted via containers. Same product, your choice of boundary.

Data stays protected

Code, context and data stay within your own infrastructure. Encrypted by default, with managed keys or your own.

Control over AI usage

Use your own approved models and AI providers. See exactly how many tokens every scan uses.

If context is the new code, that context must be under the same control.

Plans

Plans that scale with your organization.

Start with living documentation, add AI agents and change insight, or scale up with Enterprise and host FlowDex yourself.

Foundation

Living documentation and insight for your whole landscape.

Organizations up to 100 people

Talk to us
  • FlowDex web app and API
  • Living documentation: flows, focus areas and business rules
  • Code, infrastructure as code and configuration
  • Security and quality insight
  • CVEs, secrets, licenses and SBOM
  • Complexity, test coverage and duplicated code
  • GitHub, Azure DevOps, GitLab and Bitbucket
  • Bring your own AI model
  • Single sign-on
  • Hosted as SaaS
  • Basic support
Recommended

Co-Creation

For teams that build with AI agents every day.

Organizations up to 100 people

Book a demo
  • Everything in Foundation
  • FlowDex MCP for Copilot, Claude and Cursor
  • Agent workflows: build features, fix issues, close test gaps
  • FlowDex DIFF
  • Change impact and retest advice per pull request
  • Quality gate in your CI/CD pipeline
  • Feature and story generation
  • In Azure DevOps and Jira

Enterprise

For larger organizations that need scale and full control over their data.

Tailored to your organization

Contact sales
  • Everything in Co-Creation
  • Self-hosting as an option
  • Private cloud or self-hosted containers
  • Bring your own encryption key
  • Tenants, roles and audit trail

Return on investment

See what FlowDex saves your team.

At €75 an hour, one hour saved per engineer per week is already worth about €290 a month. And that is before you count faster delivery.

Calculate your savings

Custom work and consultancy available on request.

FlowDex at a glance

What it is
Code intelligence platform for AI-driven software development
For
Developers, product owners, AI agents and engineering leadership
Deployment
SaaS, private cloud or self-hosted containers
AI models
Bring your own: OpenAI, Azure OpenAI, Anthropic, Gemini, Ollama
Languages
.NET (C#) and TypeScript / JavaScript
Integrations
GitHub, Azure DevOps, GitLab, Bitbucket, Jira, CI/CD, MCP
Plans
Foundation, Co-Creation and Enterprise
Made by
Brainstack, the Netherlands

FAQ

Questions, answered.

What is FlowDex?

FlowDex is a code intelligence platform for AI-driven software development. It indexes your source code, infrastructure and configuration and turns it into living documentation: end-to-end flows, focus areas, business rules, dependencies, and quality and security findings. Developers, product owners and AI agents use that knowledge through the FlowDex app, API and MCP server.

How is FlowDex different from GitHub Copilot, Cursor or Claude?

GitHub Copilot, Cursor and Claude are AI coding assistants: they write code. FlowDex is the context and control layer underneath. It tells them — and you — how the system actually works, what a change affects and what needs to be retested. FlowDex works with those assistants through MCP instead of replacing them.

Does our source code leave our infrastructure?

Not if you don't want it to. FlowDex runs as SaaS, in your private cloud or fully self-hosted in containers. Code, context and data can stay in your own infrastructure, encrypted by default, with the option to bring your own encryption key.

Which AI models does FlowDex use?

Your own. You bring an approved provider such as OpenAI, Azure OpenAI, Anthropic, Google Gemini, Ollama or a custom LLM. FlowDex only re-synthesizes flows that changed since the last scan, which keeps token usage low.

Which languages and platforms are supported?

FlowDex analyzes .NET (C#) and TypeScript/JavaScript codebases, including frontends and message-based architectures such as Azure Service Bus, MassTransit and NServiceBus. It connects to GitHub, Azure DevOps, GitLab and Bitbucket. More languages are on the roadmap.

What is the FlowDex MCP server?

MCP (Model Context Protocol) is the open standard AI agents use to call tools. The FlowDex MCP server gives agents tools such as focus_areas, flow_insight, find_elements, impact, affected and quality_retest_advice, so they reuse existing code, check impact before editing and pick the smallest safe test scope.

Can FlowDex build features and fix security issues with my AI agent?

Yes. FlowDex MCP gives AI agents fixed workflows. /flowdex_implement turns a feature into stories, writes tests first, checks reuse and impact before the first edit, builds the smallest slice and compares the result. /flowdex_fix_issues fixes security and quality findings, such as vulnerable packages or overly complex code, one at a time. /flowdex_test_coverage adds tests for the riskiest untested code. Every run uses an isolated branch, runs checks before and after the AI edit, and ends in a pull request with evidence.

How does FlowDex prevent AI hallucinations in its documentation?

FlowDex first builds a deterministic graph of your code: entrypoints, calls, messages and data access. AI only summarizes that evidence. Every claim links back to the code, and anything FlowDex cannot prove is shown as an explicit gap instead of being invented.

Which plans does FlowDex offer?

Three. Foundation gives you living documentation and security and quality insight. Co-Creation adds FlowDex MCP, agent workflows and FlowDex DIFF. Both plans are for organizations of up to 100 people. Enterprise adds the option to self-host FlowDex in your private cloud or in containers. Book a demo and we will put together a proposal that fits your organization.

Who builds FlowDex?

FlowDex is built by Brainstack, a Netherlands-based software engineering company specialised in cloud-native and AI solutions.

Ready to experience the impact?

See FlowDex in a live demo, or start a proof of concept on your own codebase. Code. Clarity. Control.

Book a demo Read the business case

Or email us at info@flowdex.dev