AI-SHIPR · Reference Guide

AI-SHIPR at Every Stage of Your Product

Which agents, skills, and workflows belong to each stage of the product lifecycle — from first idea to post-launch iteration.

Stage 1 Discovery

The stage most teams skip. The stage that determines everything.
Every assumption you ship without validating becomes a bug you will fix in production — at ten times the cost. AI-SHIPR in discovery mode treats every bet as an unvalidated hypothesis and surfaces your highest-risk assumptions before you write a line of spec.

Settings.md product_stage: discovery
What you are trying to answer
  • Is the problem real and worth solving?
  • Who exactly has this problem — and in what context?
  • What do they use today, and why would they switch?
  • What is the riskiest assumption we are making?
  • Have we tested that assumption before committing to build?
Tool Type When to run
/discovery — Discovery Sprint Workflow Start of phase (setup), start of each sprint (check-in), end of phase (/discovery gate)
Agent Type What it does
Problem Framer Agent Turns a vague idea into a specific, bounded problem statement. Run this first.
Product Strategist Agent Builds full strategy from JTBD → Opportunity Solution Tree → OKR. Orchestrates six discovery skills in sequence.
Lean Product Canvas Agent Structures the product using the 8-box Lean Canvas. Problem before solution.
# Skill When to run
1 Competitive Landscape Mapper After Problem Framer, before user research
2 Persona Builder Before interviews; refine after
3 Interview Guide Builder Before every round of user interviews
4 Interview Synthesizer Immediately after each interview batch
5 Research Synthesizer After Interview Synthesizer; before JTBD Mapper
6 JTBD Mapper After research synthesis
7 Opportunity Tree After JTBD Mapper
8 Assumption Extractor After Opportunity Tree; before Hypothesis Builder
9 Hypothesis Builder After Assumption Extractor
10 Experiment Designer After Hypothesis Builder
11 Discovery Readiness Checker End of each discovery sprint; gate before PRD Builder
The Discovery Chain
Problem Framer
Competitive Landscape Mapper
Persona Builder
Interview Guide Builder
[interviews happen]
Interview Synthesizer
Research Synthesizer
JTBD Mapper
Opportunity Tree
Assumption Extractor
Hypothesis Builder
Experiment Designer
Discovery Readiness Checker
Exit Gate

Run /discovery gate before moving to Definition. Discovery Readiness Checker scores five categories and surfaces explicit blockers. A score below 80 or any blocked category means the move is premature.


Stage 2 Definition

Discovery is done. Now define exactly what you are building and why.
Definition converts validated hypotheses into requirements, priorities, and a roadmap. This is where PRDs get written, OKRs get locked, and the build team gets aligned before a single sprint starts.

Settings.md product_stage: definition
Tool Type What it does
PRD Builder Skill Converts a validated initiative + hypotheses into a PRD. Requires at least one confirmed hypothesis.
Roadmap Planner /plan Workflow Maps strategic bets to sequenced initiatives. Surfaces coverage gaps and sequencing risks.
OKR Partner Skill Sets objectives and key results with specificity and explicit bet linkage.
Decision Architect Agent Structures complex decisions with competing options. Run when the path is unclear.
Priority Stack Skill Scores and ranks initiatives against each other before committing to a roadmap.
Stakeholder Translator Agent Prepares for alignment meetings. Translates structured strategy into stakeholder language.

Stage 3 Building

You are in active development sprints. Keep the bet in view.
The risk in the build stage is scope creep and losing the connection between what is being built and the hypothesis it is supposed to test. AI-SHIPR keeps the link between sprint work and strategic bets explicit at every ceremony.

Settings.md product_stage: building
Tool Type What it does
Sprint Planner /sprint Workflow Sprint kickoff: goal tied to a strategic bet, committed scope, out-of-scope list.
Build Companion Skill Structures in-sprint PM decisions — scope changes, trade-offs, mid-sprint discoveries.
Build Review Skill Prepares the sprint review. Shipped vs committed. Hypothesis coverage check.
Retro Facilitator Skill Structures the sprint retro into actionable process improvements.
Fire Responder Skill Classifies and responds to active incidents. Keeps the sprint from derailing.

Stage 4 Launched

The product is live. Now you learn whether the hypotheses were right.
Launch is not the end — it is the start of the measurement window. Every hypothesis filed during discovery now has a result to check against. The question is not "how did the launch go?" but "what did the data tell us about the bet we made?"

Settings.md product_stage: launched
Tool Type What it does
Launch Coordinator /launch Workflow Pre-launch go/no-go checklist. Run 3–5 days before any launch.
Performance Tracker Skill Validates or invalidates the hypothesis with post-launch data. Returns a verdict, not a summary.
Sprint Summarizer Skill Generates sprint summaries in two modes: internal team and external stakeholders.
Weekly Review /weekly Workflow Friday loop close. What moved, decided, learned. Updates Learning.md.

Stage 5 Iterating

Data changed your direction. Now act on it — systematically.
Iteration is not "fix the bugs and add the features." It is a new hypothesis cycle, informed by what the data showed. The Iteration Planner closes the previous bet and opens the next one with explicit reasoning for why the direction changed.

Settings.md product_stage: iterating
Tool Type What it does
Iteration Planner /iterate Workflow Closes one cycle and opens the next. Run after a hypothesis is validated or invalidated.
Experiment Designer Skill Designs the next test based on what the data showed. Iteration starts with a new experiment.
Hypothesis Builder Skill Reformulates the hypothesis based on what was learned. Every iteration starts with an updated hypothesis.
Opportunity Tree Skill Re-ranks opportunities based on post-launch signal. What looked risky in discovery may now be clear.

One line in Settings.md changes how every agent and workflow behaves. Update it deliberately — each transition is a strategic decision, not a calendar event.

Stage Transitions
Discovery → Definition
Discovery Readiness Checker scores 80+ with no blocked categories
Set product_stage: definition — run PRD Builder on highest-confidence bet
Definition → Building
PRD approved, OKRs locked, team aligned, no open blocking decisions
Set product_stage: building — run Sprint Planner for sprint 1
Building → Launched
Launch Coordinator checklist passes go/no-go
Set product_stage: launched — start measurement window
Launched → Iterating
Performance Tracker returns a verdict — confirmed, invalidated, or inconclusive
Set product_stage: iterating — run Iteration Planner to open next cycle