For controlled-environment modernization

Modernize legacy code with local AI inside infrastructure you control.

Mirepoix brings AI-assisted understanding, porting, review, and validation into customer-controlled infrastructure. Its sovereign runtime profile requires a loopback model provider, workspace-confined tools, denied tool egress, and a replayable JSONL record. Host firewall, IAM, and deployment evidence remain independently reviewable.

The job Legacy modernization
Where it runs Your infrastructure
Runtime controls Fail closed, auditable
Engagement Pilot and partner delivery
A new category of modernization tooling

Local AI assistance for modernization inside controlled infrastructure.

Teams often have to choose between conventional tooling inside a controlled environment and hosted AI outside it. Mirepoix is designed to close that gap with local-model port assistance, multi-agent review, equivalence validation, and evidence from both the runtime and its deployment.

Modernization without this tooling

What's been available until now

Modernization is delivered by partners with deep engineering experience and existing customer relationships. The work happens with the tooling the market has had — which has not, until now, included AI assistance that can operate inside the customer's perimeter.

  • AI assistance available only on hosted platforms that cannot operate inside the perimeter
  • Limited evidence showing how model traffic and tool access were constrained during a run
  • Multi-agent review and equivalence validation not part of standard modernization tooling
  • Audit trails captured at workflow level, not at the AI tool-call level
  • Engagement timelines constrained by what's possible without modern AI-driven port assistance
  • Hosted-AI tools may not fit workloads that prohibit sending source to an external model provider
Modernization with Mirepoix · Local-first

The new tooling layer your partners can now bring

Mirepoix runs in customer-controlled infrastructure. Your engineers or delivery partner operate the engagement, with the model provider constrained to loopback and tool access constrained by the declared runtime profile.

  • Deployable software and documented operating procedures
  • AI-driven port assistance using a locally served model
  • Multi-agent face-off review catching errors before they ship
  • Equivalence validation between legacy and modernized code, automated
  • Fail-closed runtime controls, with deployment evidence reviewed separately
  • Replayable JSONL event records for agent and tool activity
How local modernization works

A five-stage pipeline operated inside your environment.

Mirepoix can run the modernization pipeline with a locally served model in customer-controlled infrastructure. Under the sovereign runtime profile, model traffic is restricted to loopback, file writes stay inside the workspace, and tool network access is denied. Outputs remain reviewable and validation compares the port with the legacy behavior.

01 / UNDERSTAND

Map the codebase

Architecture graph, dependency-ordered tour, and per-module summaries built from the source in the controlled workspace.

02 / PORT

Module-by-module

Generate the modernized port to the target language (Rust, Go, modern C++, TypeScript) following the dependency tour.

03 / REVIEW

Multi-agent face-off

Two independent reviewer agents adjudicate each port. Hallucinations and integration errors are caught before they ship.

04 / VALIDATE

Equivalence evidence

Generated test suites compare modernized code against legacy behavior. Numerical equivalence is measured to an explicit tolerance where applicable.

05 / AUDIT

Replayable record

JSONL event records capture agent, tool, review, and validation activity for replay and investigation.

What Mirepoix can demonstrate today

Runtime controls and deployment evidence, reviewed separately.

Mirepoix makes bounded claims about the controls it enforces. The harness can fail closed on model routing and tool access. The deployment can separately collect evidence for firewall, IAM, and service configuration. Current GPU deployments do not produce Intel TDX or AMD SEV-SNP hardware-attestation quotes.

  • Local model containment The sovereign runtime profile accepts only a loopback model provider and refuses HTTP redirects.
  • Workspace write confinement Typed write and edit operations are restricted to the workspace, including defenses against symlink escapes.
  • Tool egress denial Sovereign mode refuses tool-network exceptions and requires an available process-sandbox mechanism before a session starts.
  • Independent deployment evidence Host firewall and IAM evidence is collected and reviewed outside the harness. It is not represented as hardware attestation.
sovereign runtime profile
provider : loopback only
redirects : refused
writes : workspace only
tool_network : denied
process_sandbox: required at startup
session_record : JSONL events
host_evidence : reviewed separately
DEPLOYABLE

Software for controlled infrastructure

A reusable harness, modernization packages, and operating runbooks that can be evaluated in a bounded pilot.

LINUX

Customer-controlled deployment target

The current validated posture uses a GCP GPU VM and a local model. Other platforms require their own deployment qualification.

BYO MODEL

Your model, your weights

Choose and operate an open-weight code model locally. In sovereign mode, Mirepoix refuses non-loopback provider URLs and redirects.

OPERABLE

Documented controls

Runbooks, acceptance gates, and explicit responsibility boundaries make the operating posture reviewable.

For cleared contractors and federal primes

Bring the workforce. Bring the relationship. We bring the platform.

The partnership structure

Delivery partners bring customer relationships and engineering capacity. Mirepoix supplies a local-first harness for AI-assisted understanding, porting, review, and equivalence validation, with runtime controls that can be inspected before an engagement begins.

The partner operates Mirepoix inside customer-controlled infrastructure under the customer's security and authorization process. Model selection, network policy, operator access, and acceptance criteria remain explicit deployment decisions.

This structure is intended for organizations evaluating whether local AI assistance can fit their controlled-development workflow. Any regulatory, contractual, or authorization claim must be assessed for the specific customer deployment.

Engagement responsibilities
Contractor (federal prime, cleared consultancy)
Cleared engineering workforce. Existing agency relationships and contract vehicles. Customer-facing engagement architecture and delivery. ATO ownership.
Mirepoix
Software, modernization methodology, operating documentation, technical support, and platform updates.
Agency or commercial customer
Deployment infrastructure, operator access, model weights, network controls, review of deployment evidence, and acceptance gates on modernization output.
Measured, not extrapolated

Current evidence from bounded validation work.

Mirepoix's founding engagement is a Python-to-Rust modernization using a locally served model on a GCP A100 host with a deny-all public-egress posture. Repository tests exercise runtime confinement and equivalence behavior. These results do not establish hardware attestation, regulatory compliance, or universal performance.

Equivalence precision
10decimal places

The validation package supports explicit absolute and relative tolerances; an internal Kirk-precursor pattern compared numerical output to about ten decimal places.

Runtime regression suite
72focused tests

Focused tests cover sovereign configuration, loopback model routing, redirect refusal, process-sandbox startup checks, and workspace confinement.

Repository verification
695tests passing

The full repository test run passes with zero failures, alongside package type checks, formatting checks, and ADR consistency checks.

Evaluate local AI modernization on a representative module.

Bring a bounded legacy module and its expected behavior. We can demonstrate codebase mapping, a modernization pass, multi-agent review, equivalence validation, and the runtime evidence produced along the way.

LOCAL MODEL · WORKSPACE CONFINEMENT · REPLAYABLE AUDIT