Doddle Conversion OS — 30 agent-agnostic skills and 5 agents for conversion rate optimization: strategy, research, landing pages, offers, experimentation, personalization and measurement.
The one claim this pack makes: every number is reproducible from recorded evidence. A
stdlib-only Python toolkit (dconv) computes; the skills narrate and prioritize. Baseline
rates, MDEs and guardrails are assumptions and are always labelled advisory.
Point your agent at .claude/skills/ and .claude/agents/, or install the DoddleOS plugin
from .claude-plugin/plugin.json.
python3 scripts/dconv doctor
python3 scripts/dconv versionPython 3.9+, no dependencies (stdlib only).
dconv version
dconv frameworks [--name SLUG] [--json]
dconv ab-test --control-n N --control-conv C --variant-n N --variant-conv C [--json]
dconv sample-size --baseline R --mde D [--alpha A] [--power P] [--json]
dconv funnel --stage name:count ... [--json]
dconv landing-audit --checks '{...}' [--file F] [--json]
dconv offer --price P [--discount D] [--bundle JSON] [--json]
dconv doctor
Skills (30) grouped by the conversion motion:
| Group | Skills |
|---|---|
| Strategy & research | conv-strategy, cro-audit, cro-research, cro-heuristics, cro-copy, cro-ux |
| Landing pages | conv-landing-page, conv-landing-audit, cro-forms, cro-mobile, cro-speed, cro-trust |
| Offers & pricing | conv-offer-design, conv-pricing-page, conv-discount-strategy, conv-bundle-design, conv-guarantee |
| Experimentation | conv-experiment-design, conv-ab-testing, conv-multivariate, conv-hypothesis, conv-significance, conv-experiment-reporting |
| Personalization | conv-personalization, conv-segmentation, conv-dynamic-content, conv-geo-targeting |
| Measurement | conv-metrics, conv-funnel, conv-attribution |
Agents (5): conversion-strategist, cro-analyst, ux-researcher, experiment-lead,
offer-strategist.
Full inventory: skills-matrix.md. Authoring rules: AGENTS.md.
python3 scripts/lint_pack.py && python3 -m pytest tests -qMIT. See LICENSE.