{"name":"@crewd/experiment-design","version":"1.0.0","type":"skill","description":"Design experiments with a clear hypothesis, metric and stopping rule.","integrity":"sha256-b68c944d45742f3939a8fba847cb54213ec1a2d7dba042b5cfbd9fbac040e0ab","tarball_url":"https://localhost:3000/api/v1/tarballs/crewd/experiment-design/1.0.0","tarballUrl":"https://localhost:3000/api/v1/tarballs/crewd/experiment-design/1.0.0","size_bytes":311,"hosts":["claude-code","cursor","gemini-cli"],"manifest":{"name":"@crewd/experiment-design","type":"skill","entry":"SKILL.md","hosts":["claude-code","cursor","gemini-cli"],"schema":"https://crewd.dev/schema/skill/v1","content":"---\nname: experiment-design\ndescription: \"Design experiments with a clear hypothesis, metric and stopping rule.\"\n---\n\n# Experiment Design\n\nUse this skill before running any A/B test or experiment.\n\n## Template\n\n- **Hypothesis**: If we do X, then Y will change, because Z.\n- **Primary metric**: one metric, defined precisely.\n- **Guardrail metrics**: things that must not get worse.\n- **Sample size**: computed from baseline, minimum detectable effect, power 80%, alpha 5%.\n- **Duration**: at least one full weekly cycle.\n- **Decision rule**: ship / iterate / drop, decided before launch.\n\n## Pitfalls\n\n- Peeking and stopping early.\n- Changing several things at once.\n- Testing on too small an audience.\n- Ignoring novelty effects.\n","version":"1.0.0","description":"Design experiments with a clear hypothesis, metric and stopping rule."},"scripts":[],"dependencies":[],"deprecated":null,"created_at":"2026-10-18T17:02:37.717758+00:00"}