bootstrap

name	bootstrap
bootstrap-phase	product
description	Bootstrap the product analytics documentation structure for a project. Creates docs/analytics/, generates initial metric-tree and instrumentation templates, and writes the product-analyst fragment of the analytics domain doc (the coordinator assembles docs/analytics/CLAUDE.md). Idempotent — merges missing sections into existing files without overwriting.
argument-hint	[project name]
user-invocable	false
allowed-tools	Read, Write, Edit, Bash, Glob, Grep

Bootstrap Product Analytics Documentation

Bootstrap the product analytics documentation structure for $ARGUMENTS. This sets up the home for North Star definitions, metric trees, instrumentation specs, cohort analyses, and experiment designs produced by the other product-analyst skills.

Step 1: Check and create domain directory

mkdir -p docs/analytics docs/analytics/_sections

Step 2: Create or merge files

For each file below, apply the safe merge pattern:

• If the file does not exist → create from template
• If the file exists → read both, find sections in the template missing from the file, append the missing sections with <!-- Merged from product-analyst bootstrap v0.1.0 -->

Never overwrite existing content. Bootstrap is idempotent — running it twice changes nothing the second time.

File 1: docs/analytics/_sections/product-analyst.md

docs/analytics/CLAUDE.md is assembled by the coordinator from the fragments in _sections/ — no plugin writes it directly. Write the product-analyst's contribution as this fragment. It starts at H2 (the coordinator generates the # Analytics Domain H1 and a one-line intro). Create it with this content:

## What this domain covers

- **North Star Metric** — the single number capturing delivered customer value
- **Metric hierarchy** — input metrics that drive the North Star (HEART or AARRR)
- **Instrumentation specs** — events, properties, identity model, attribution (handed to data-engineer)
- **Cohort and retention analysis** — retention curves and segment cuts
- **Experiment design** — A/B and holdout tests with sample-size calculations

## Ownership boundary

The product-analyst defines *what* to measure and *why*. The data-engineer builds *how* data
flows (pipelines, warehouse, dashboards). Instrumentation specs are written here and handed off
for implementation. OKRs live with the product-manager; metrics here serve those OKRs.

## Metric definition standard

Every metric carries a definition block before implementation:

- **Question it answers** — the decision the number informs
- **Definition** — precise and unambiguous
- **Calculation** — exact formula or query logic
- **Granularity** — per user / session / account / cohort
- **Filters** — bots, test accounts, internal users (explicit)
- **Time window** — rolling 7d / calendar month / since signup
- **Goodhart risk** — how it could be gamed and who it would hurt
- **Owner** — who approves a definition change

## Conventions

- The North Star measures customer value, never company revenue directly
- Run the Goodhart check on every metric before promoting it
- Vanity metrics (cumulative totals) are never a North Star or an input
 Correlation is the default; causation requires an experiment
 Evidence hierarchy: observed behaviour > stated intent > opinion
 One file per artifact under 



| Skill | Purpose |
|-------|---------|
|  | Define the North Star Metric and its input metrics |
|  | Build a HEART or AARRR metric tree tied to OKRs |
|  | Specify events for the data-engineer to implement |
|  | Analyse retention curves and segment cuts |
|  | Design an A/B or holdout experiment |

File 2: docs/analytics/metric-tree.md

Create from the North Star canvas and metric-tree templates in the plugin's templates/ directory (north-star-canvas.md, metric-tree.md). Copy the template body, leaving the placeholder rows for the team to fill via /product-analyst:define-north-star.

File 3: docs/analytics/instrumentation-spec.md

Create from the plugin's templates/instrumentation-spec.md. This is the hand-off artifact to the data-engineer; leave the event table empty for /product-analyst:write-instrumentation-spec to populate.

Step 3: Return manifest

After creating or merging all files, output a summary:

## Product Analytics Bootstrap Complete

### Files created
- `docs/analytics/_sections/product-analyst.md` — product-analyst fragment (coordinator assembles `docs/analytics/CLAUDE.md` from it)
- `docs/analytics/metric-tree.md` — North Star and metric-hierarchy template
- `docs/analytics/instrumentation-spec.md` — data-engineer hand-off template

### Files merged
- (list any existing files where sections were appended)

### Next steps
- Define the North Star with `/product-analyst:define-north-star`
- Build the metric hierarchy with `/product-analyst:design-metric-hierarchy`
- Write the instrumentation spec and hand it to the data-engineer

Rules

• Always create docs/analytics/, never docs/product-analytics/ or a variant — one canonical path
• Never overwrite an existing file. Merge missing sections only, tagged with the merge comment
• Do not invent metric values during bootstrap — templates ship with empty placeholder rows
• Don't duplicate OKR content here — OKRs live with the product-manager; link, don't copy

Output Format

Return the manifest from Step 3 verbatim: files created, files merged, next steps. Nothing else.