Platform Workflow

How Feature1 Works

From product idea to deployed code — powered by domain intelligence and AI agents

The Feature1 Pipeline

Six stages from repo connection to shipped code

Connect Your Repository

Onboard your GitHub, GitLab, or Bitbucket repository in minutes. Feature1 securely indexes your codebase and establishes a persistent connection to track changes over time.

GitHubGitLabBitbucketOAuth 2.0

Build Domain Intelligence

The AI constructs a semantic knowledge graph of your codebase — understanding modules, dependencies, conventions, and architectural patterns. This graph powers every downstream decision.

Knowledge GraphSemantic AnalysisAST ParsingDependency Mapping

Plan Features

Submit a feature idea in plain language. Feature1 runs a 6-stage AI analysis pipeline — scoping, risk assessment, dependency analysis, effort estimation, acceptance criteria drafting, and implementation planning.

6-Stage PipelineRisk AnalysisEffort EstimationScope Definition

Generate User Stories & ACs

User stories and acceptance criteria are auto-generated from feature analysis, fully aligned with your domain model. Each story is traceable, testable, and ready for implementation.

User StoriesAcceptance CriteriaBDD FormatTraceability

Implement with AI

Choose Autopilot — AI implements each AC autonomously, commits code, and opens PRs with approval gates — or Copilot for Driver-Navigator pair programming via your MCP client (Claude Code or Codex CLI).

AutopilotCopilotContext-AwareStyle Consistent

Ship

A pull request is created automatically with a full diff, inline comments explaining decisions, and auto-generated release notes. Merge when ready — Feature1 handles the rest.

Auto PRRelease NotesInline DocsCI/CD Ready

Choose Your Mode

Two ways to work with Feature1 — pick the level of control that suits your team

Autopilot

HITL Autopilot

Feature1 automates the heavy lifting between approval gates. AI plans, generates user stories and acceptance criteria, implements code, creates PRs, and drafts release notes, while humans approve critical transitions.

Automation between human approval gates
Feature analysis to PR with controlled checkpoints
Auto-generated release notes on every merge
Continuous learning from merged code
Ideal for well-scoped, repeatable work

Best for: Teams who want maximum velocity with minimal overhead

Copilot

Driver-Navigator

The MCP client (Claude Code or Codex CLI) drives the implementation while you navigate — guiding design decisions and architecture choices at each step.

Driver-Navigator pair programming pattern
MCP client (Claude Code or Codex CLI) drives code
You navigate design decisions and architecture
Guide each step without writing it yourself
Full audit trail of every AI action

Best for: Teams who want AI leverage while keeping architectural ownership

MCP Integration

Connect Claude Code and other AI agents directly to Feature1 via the Model Context Protocol

What is MCP?

The Model Context Protocol (MCP) is an open standard that lets AI agents securely connect to external tools and data sources. Feature1 exposes its entire workflow through an MCP server — meaning any compatible agent can plan, implement, and ship features programmatically.

Claude Code

Cursor

Windsurf

Custom Agents

feature1-mcp-server

create_feature()// submit a new feature idea

get_planning_context()// fetch codebase intelligence

list_user_stories()// retrieve generated stories

implement_ac()// trigger AI code generation

mark_ready_for_testing()// flag AC for QA

create_pr()// open pull request with release notes

Secure by Default

OAuth-scoped tokens, no plain-text secrets

Real-Time Context

Always-fresh codebase knowledge graph

Bidirectional

Read and write across the full workflow

Agent Agnostic

Works with any MCP-compatible client

Workflow State

Track progress across every pipeline stage

REST Fallback

Standard API available alongside MCP

Ready to ship faster?

Connect your repo, let Feature1 build domain intelligence, and start delivering features at AI speed.