Product Strategy Canvas pack. product-strategy-canvas (this skill). Upstream: user-research, opportunity-finder, strategy-sprint. Downstream: prd-writer, backlog-builder, experiment-design, insight-storyteller.

Product Strategy Canvas

Product teams have plenty of artifacts: interview notes, opportunity lists, roadmaps and KPI decks. What they lack is the chain that connects them. This skill builds that chain in one file and draws it, so anyone can click a roadmap item and see the bet, the opportunity, the customer need and the evidence behind it.

Your role: keep the canvas honest and connected. Draft, score and draw; the team decides.

Ground rules

• One source of truth: canvas-output/canvas.json (schema in references/canvas-schema.md). Every step reads it and updates it. Never retype earlier outputs.
• Everything has an ID and a link upstream: S source → E evidence → N need → O opportunity → B bet → I initiative → K KPI. An item with no link upstream is an orphan, and orphans get flagged.
• Labels (references/evidence-rules.md): [FACT S#], [ESTIMATE S#], [INFERENCE], [HYPOTHESIS], [UNKNOWN]. Bets are always hypotheses until the scorecard says otherwise.
• No invented evidence: no made-up quotes, survey numbers, reach figures or baselines. Unknown baselines stay [UNKNOWN] with a plan to measure them.
• Read-only default: write only to canvas-output/. Creating tickets, sending messages or publishing needs explicit approval.

Step 0 — Mode and starting point

If canvas-output/canvas.json exists, run scripts/canvas_check.py on it, summarize where it stands, and continue.

Otherwise pick a mode:

Reuse what the kit already has instead of redoing it:

• Opportunity Finder output (opportunity-output/): import the evidence, needs and top opportunities at steps 01–03.
• Strategy Sprint decision (strategy-output/): import the choices and approved option at step 04.
• User Research synthesis: import themes as evidence at step 01.

Copy assets/templates/canvas-blank.json to canvas-output/canvas.json and fill meta (product, objective, owner, as_of, mode, prioritization method).

Steps (load one playbook at a time: references/steps/NN-<step>.md)

After each step: run python scripts/canvas_check.py canvas-output/canvas.json. Fix errors before moving on, and report gaps to the user.

Render

python scripts/render_canvas.py canvas-output/canvas.json -o canvas-output/

This produces:

• canvas.html: one interactive page. Click any card to light up its full chain.
• Five SVGs for decks and docs: opportunity tree, prioritization matrix, roadmap, dependencies, KPI tree.

Share the HTML as an artifact when the user wants a page to keep or share.

Answer "why is this on the roadmap?"

python scripts/canvas_check.py canvas-output/canvas.json --why I3

This prints the chain from initiative to bet to opportunity to need to evidence to source, plus the success metric and kill criteria. Use it whenever someone challenges a roadmap item.

Export

python scripts/export.py canvas-output/canvas.json -o canvas-output/

This produces three files:

• backlog.csv: epics per bet and stories per initiative, for Jira or Linear. Refine the stories with backlog-builder.
• prd-inputs/B#.md: one PRD starter per bet, for prd-writer.
• deck-brief.md: the action-title spine for a leadership readout.

Scripts (Python 3, standard library only)

Templates (assets/templates/)

canvas-blank.json, canvas.json (filled fictional example), bet-card.md, scorecard.md, review-cadence.md.

A fictional end-to-end run is in examples/worked-example.md, with rendered outputs in examples/sample-output/. Test prompts are in evals/evals.json.

Credits

Opportunity Solution Tree (Teresa Torres), RICE (Intercom), ICE (Sean Ellis), WSJF (Don Reinertsen; SAFe), outcome-based roadmaps (Josh Seiden, Jeff Gothelf), Now/Next/Later (Janna Bastow), North Star framework (Amplitude), Outcome-Driven Innovation (Tony Ulwick).