MCP software · Private pilot
PRAE Platform
Run security checks on the project in front of you. From your coding assistant or command line, inspect findings, verify fixes and export the evidence.
How it works
Your project sets the scope.
Choose the inputs
Approve the files, build and environment the local engine may examine. Supported checks cover npm dependencies, GitHub Actions, declared agent permissions and selected hosting settings.
Ask your assistant to run a check
The Model Context Protocol (MCP) connects your assistant to 21 local tools for assessment, explanation, verification and reporting. The assistant works within the scope you approved. Expanding it requires a human decision.
Retest and keep the evidence
Make the fix in your normal development workflow. A fresh assessment records the result beside the original finding. An operator decides whether to close it. Export the record as JSON, Markdown or HTML.
From a synthetic pilot assessment
A dependency declared in one file disagreed with another.
The npm manifest and lockfile gave different accounts of a direct dependency. The team corrected the fixture and ran the check again. The report retained the original mismatch and recorded the finding as resolved with evidence.
- Scope
- Four approved files examined
- Final run
- Six assertions passed. Six were recorded as inapplicable with reasons.
- What the result establishes
- The corrected declarations agree. Confirming which version is deployed requires separate evidence.
Two surfaces, one project
Work through your agent.
Review the evidence here.
MCP · the primary workflow
The local or private runner discovers supported infrastructure, assesses approved files, traces modeled data journeys, prepares remediation and verifies bounded changes. Your agent returns links to private visual artifacts.
The hosted MCP connection provides ten review tools for shared project health, maps, data-flow evidence and the artifact library. It cannot run code in a remote repository or collect an unseen production system.
Workspace · identity and oversight
Use the browser to sign in, choose your project, review schematics and evidence, manage individual connections and revoke access. A team owner assigns named people to specific projects. Billing and signed downloads become available when commercial release gates pass.
Personal access, explicit limits
One individual subscription covers one named person and one project. The Team plan covers five named people and five projects across the workspace. Manual credentials are individual, project-bound and expire within seven days; removing access revokes online permissions. Local offline receipts remain valid only until their stated expiry.
Access and pricing
Annual plans
Individual project
US$399 per year · 1 person · 1 project
Team
US$1,999 per year · 5 named people · 5 projects
Prices exclude applicable tax, which is calculated and shown before payment.
The software remains in private pilot. The workspace supports named accounts, governed projects and private evidence views. Individual covers one person and one application. Team covers five named people and five applications, including their development, staging and production environments. Larger teams can request a scoped quote. Paid checkout remains closed while client acceptance, signed distribution and commercial terms are completed.