Skip to main content

Polygent

Polygent is a self-hosted control plane for running and supervising AI-assisted development across shared Git repositories.

A deployment consists of the web application, a database, and session execution capacity. Optional deployment workers host long-lived preview environments. Each development session uses an isolated Git worktree so concurrent tasks do not share a working directory.

Core capabilities

Polygent groups day-to-day development and operations into these areas.

AreaCapability
SessionsRun Chat, Develop, and Bot sessions against selected models in isolated worktrees.
TicketsTrack implementation, developer review, QA, pull requests, failures, and completion.
AutomationReuse workflows, bots, scheduled automations, planning, and round-table reviews.
GitReview changes, commit and push branches, create pull requests, and resolve merge conflicts.
OperationsManage workers, preview deployment slots, logs, budgets, notifications, licenses, and branding.
GovernanceRestrict actions with roles, permissions, workspace membership, and authenticated worker credentials.
ReportingReview session, ticket, cost, automation, and user-activity statistics.

Deployment and data boundaries

Polygent runs on infrastructure you control and supports SQLite, PostgreSQL, and SQL Server. Repository data, session records, credentials, logs, and uploaded files remain within the systems you configure, except where an enabled integration sends data to an external service. Hosted models and ticket providers receive the data required for their configured operations; use local models when model traffic must not leave your network.

Secure production deployments with HTTPS, an OAuth or OpenID Connect identity provider, least-privilege roles, restricted worker credentials, protected storage, and tested database and storage backups.

Use this order for a new production deployment.

  1. Installation — install the application and optional workers.
  2. System Requirements — size hosts and verify dependencies.
  3. Architecture & Security — review trust boundaries and hardening controls.
  4. Quick Start — configure a workspace and run a session.
  5. Permissions — define roles and workspace access before onboarding users.