Skip to content
floe studio.
Available now

PAAP

Your AI product team in a terminal.

Product-Aligned Agent Process. 9 specialized agents, a reactive automation layer, and Agent Teams — covering the full product lifecycle from challenging an idea through shipping and measuring outcomes.

9Agents
13Skills
12Commands
14Templates

Product work is mostly manual.

  • —Writing PRDs from scratch every sprint
  • —Copying data between Jira, Confluence, and Figma
  • —Reviewing code without consistent standards
  • —Running discovery and delivery as separate tracks
  • —Measuring outcomes weeks after launch — or not at all
  • —Triggering reports and checks manually every week

PAAP replaces the manual parts with agents that think, act, verify — and now automate on schedule.

Four phases. One continuous loop.

01

Ideate

Challenge ideas, run discovery, map opportunities.

Product StrategyDesign
02

Validate

Test hypotheses, check feasibility, make go/no-go decisions.

Hypotheses
03

Execute

Build incrementally from validated hypotheses.

ArchitectureBackendFrontend
04

Measure

Track outcomes, feed learnings back to discovery.

Metrics
Learnings feed back from Measure to Ideate — continuous discovery & delivery

9 agents. Each one a specialist.

Every agent has identity, perspective, and decision-making authority. They're not prompts — they're thinking partners with intellectual foundations.

Master

Team Lead
All phases

Orchestrates via Agent Teams — spawns parallel teammates, tracks shared tasks, verifies. Schedules reports and checks via cron. Never writes code.

Product Strategy

Discovery
Ideate

Pressure-tests product ideas, challenges assumptions, and drives structured discovery. Never validates by default — interrogates.

Hypotheses

Validation
Validate

Manages the full hypothesis lifecycle. Adversarial — kills weak hypotheses before they waste engineering time. Cron-monitored validation status.

Architecture

Quality Gate
Execute

Reviews all code, enforces standards, makes technical decisions. Auto-lints via hooks, blocks dangerous commands. Nothing ships without approval.

Backend

Server
Execute

APIs, database schemas, business logic, integrations. Worktree isolation for parallel development. Clean, tested, production-ready code.

Frontend

UI
Execute

Components, pages, styling, interactions. File-change hooks auto-trigger design verification. Accessible, responsive, tested.

Design

UX
Ideate

User flows, information architecture, component behavior. Advises — the Frontend Agent implements.

Metrics

Outcomes
Measure

Tracks results against outcomes, generates insights, feeds learnings back. Cron-scheduled weekly reports and hypothesis outcome checks.

Prompt

Meta
All phases

Evaluates and improves prompts, instructions, and agent definitions. Quality gate for human-to-AI communication.

Slash commands that do real work.

Type a command, get a production-quality artifact. No prompt engineering required.

/hypothesis-to-prdConvert hypothesis to structured PRD
/meeting-to-actionsExtract action items from meeting notes
/sprint-reportGenerate stakeholder sprint report
/velocity-analysisAnalyze sprint velocity data
/prd-gap-checkCompare PRD vs original hypothesis
/upload-project-PRDs-ConfluenceExport PRDs to Confluence
/improve-promptEvaluate and improve prompts
/evalEvaluate artifacts against binary rubrics
/alignCheck alignment between files and tools
/researchGather context before planning
/verify-uiCompare Figma design vs implementation
/list-all-claude-commandsShow all available commands

13 skills. Domain knowledge on tap.

Skills are composable checklists and protocols that agents load on-demand. They encode best practices so the agents don't start from zero.

accessibilityExecute

WCAG 2.1 AA compliance, keyboard nav, ARIA

security-reviewExecute

OWASP Top 10, auth patterns, secrets

testing-patternsExecute

Unit / integration / E2E per layer

api-designExecute

Contract-first, validation, error handling

confluence-publishingAll phases

Storage format, table patterns, API

jira-managementAll phases

Issues, components, workflows, JQL

hypothesis-validationValidate

Lifecycle, falsifiability, Dead Zone rule

discovery-artifactsIdeate

JTBD, Value Prop, OST, Impact x Effort

delivery-workflowExecute

PRD-to-backlog pipeline, sprint patterns

metrics-measurementMeasure

OMTM, signal vs noise, feedback loop

spec-feasibility-reviewExecute

Engineering feasibility, effort estimation

stakeholder-clarity-reviewAll phases

Document scannability, jargon detection

hooks-automationAll phases

Hook patterns, cron scheduling, event catalog

One framework. One price.

All 9 agents, 13 skills, 12 commands, and every template. Agent Teams, hooks, cron scheduling included. One-time payment, lifetime access.

$20one-time
  • —All 9 agents, 13 skills, 12 commands
  • —Agent Teams, hooks, cron scheduling
  • —All 14 templates
  • —Continuous updates (repo pulls)
Get PAAP

Up and running in three steps.

01

Install

Clone the repo, run setup.sh. Agents, commands, skills, and hooks install automatically.

02

Configure

Customize intellectual foundations per agent. Connect Jira, Confluence, and Figma. Enable Agent Teams.

03

Start

"Pressure-test this idea: [your product idea]" — and the agents take it from there.

Integrates with
Claude CodeJiraConfluenceFigmaPostHog

Let’s talk product.

schedule meeting

Move from manual work to AI-powered product systems.