silver-measure
Close the loop after implementation by determining whether a shipped design or product change achieved its intended outcome — reviewing instrumentation, inspecting analytics, comparing before/after, and stating confidence and limitations. Use after implementation, or whenever an outcome needs to be checked against its original success criteria.
- Category
- Product analytics
- Status
- Experimental
- Use when
- check whether a shipped change achieved its intended outcome
- review instrumentation before trusting a metric
- compare before/after a launch
- state confidence and limitations for a measured result
- avoid treating correlation as causation in an outcome review
- Works on
- the original hypothesis and success criteria for a shipped design or product change
- available instrumentation/analytics data covering before and after the change
- Produces
- a measurement record comparing before/after or experiment results, with instrumentation sufficiency, confidence, and data limitations stated explicitly, ready to feed back into synthesis
- Requires
- node
- Works with
- Claude Code, Codex, Cursor, OpenCode, Amp, Gemini CLI, GitHub Copilot, VS Code, Claude.ai
- Tags
- analytics, instrumentation, metrics, experimentation, hypothesis
- Source
- thejparsenault/silver-design-framework by JP Arsenault · MIT
- Related skills
- silver-evaluateinstrumentation-planproduct-analyticsexperiment-designer
Instructions from SKILL.md
Measure an outcome
Workflow
- Restate the hypothesis and success criteria the change was meant to test.
- State what instrumentation exists and whether it is sufficient to answer the question.
- Compare before and after, or the experiment result, without treating correlation as causation.
- Record confidence and limitations; feed observed outcomes back into synthesis.
Run the guarded file operation through the CLI when durable outputs are ready:
.silver/bin/silver invoke --scaffold measure .
.silver/bin/silver invoke measure <request.json> .The scaffold prefills timestamps, provenance, pinned context, required checks, and
expected_integrity. Replace every silver-scaffold-placeholder before invoking;
the CLI refuses a request that still contains one.
Done
- Satisfy: metrics-tied-to-hypothesis, instrumentation-stated, data-limits-recorded.
- Evaluate quality: Observed outcomes stay traceable to the success criteria they test, with instrumentation and data limits stated rather than implied.
- Emit a valid
silver/skill-result/v2record separating execution, acceptance, and downstream readiness. - Recommend follow-up skills; never start them automatically.
Boundaries
- Do not define success criteria; that is
specify. - Do not interpret mixed research evidence broadly; that is
synthesize. - Do not fabricate analytics or claim measurement occurred without actual data.
- Do not treat correlation as causation without evidence, or overstate confidence when instrumentation or sample quality is weak.
Supporting files
4 more files in this skill's folder. Markdown opens as a page, other text files open as source, and anything else downloads as it is.