Documentation

Welcome to Pivota

Pivota is a self-hosted, AI-powered SDLC platform that orchestrates the entire software delivery lifecycle — from requirements to deployed code. Built on a Rust/Axum backend with a React/TypeScript frontend, Pivota integrates 45 slash commands and 31 specialized AI agents into a visual workflow with full traceability.

Quick StartGet productive in 5 minutes
InstallationDocker deployment guide
Core ConceptsUnderstand spec-driven dev

What is Pivota?

Pivota is a self-hosted platform that orchestrates AI-assisted software delivery with full governance and traceability. Unlike standalone AI coding tools, Pivota operates as a complete SDLC overlay:

  • Self-hosted, single Docker — Deploy as a single container. PostgreSQL + pgvector for metadata and vector search. Your code, your infrastructure.
  • 45 Slash Commands — From /pivota_spec:new-project to /pivota_spec:complete-milestone, every step of the SDLC is a command.
  • 31 Specialized AI Agents — Purpose-built agents for research, planning, execution, verification, UX design, and document generation.
  • Active Kanban Execution — Drag a card to "In Progress" and Pivota provisions a Daytona sandbox, launches an AI agent, streams output in real-time, and creates a PR on completion.
  • Full Spec Document Suite — Auto-generates PRD, FRD, TechArch, UserStories, Personas, Journeys, JTBD, StoryMap, RTM, Business Case, FIPS categorization.
  • End-to-End Traceability — From PRD to PR, every requirement maps to code. Evidence packs and audit trails generated automatically.

The Pivota Workflow

Pivota uses a spec-driven workflow with four phases:

1

Discuss

Describe what you want to build. Pivota asks clarifying questions and captures requirements in structured documents.

2

Plan

AI researches patterns, creates execution plans verified against your goals, and breaks work into atomic tasks.

3

Execute

Each task runs in an isolated git worktree with sandbox testing. Fresh context per task ensures consistent quality.

4

Verify

Automated verification against requirements. Evidence packs generated. PR created and ready for review.

Key Features

Multi-LLM Support

Claude, GPT-4, Gemini, Z.ai — switch models without changing workflows. Model profiles for quality, balanced, and budget tiers.

45 Slash Commands

Full spec-driven workflow via commands. /pivota_spec:new-project, /pivota_spec:plan-phase, /pivota_spec:execute-phase, and 42 more.

DoD P1 Assessed

Assessed Awardable on DoD Platform One Solutions Marketplace. Built for regulated, multi-team, high-compliance environments.

Kanban Execution Engine

Drag-to-execute kanban cards. Daytona sandbox provisioning. Real-time SSE streaming. Automated PR creation.

System Requirements

# Deployment
Docker 20.10+ with Docker Compose v2
PostgreSQL 16 with pgvector extension
4GB RAM minimum (8GB+ recommended)

# AI Providers (at least one required)
Anthropic (Claude) — recommended
OpenAI (GPT-4, GPT-4o)
Google (Gemini)
Z.ai

# Integrations
GitHub (primary SCM)
GitLab (supported)
Daytona Cloud (sandbox execution)
OpenCode AI Agent

Next Steps

Quick Start Guide Learn Core Concepts