The Product Skills LibraryFree AI skills for Claude & ChatGPT

Builds a discovery interview guide from one decision and its riskiest assumptions, rewrites leading or hypothetical questions, and adds a screener, probes and a notes template.

Synthesizes interview transcripts one by one, then across, counting only from a participant-by-theme matrix — findings with verified quotes, counter-evidence and capped confidence.

Turns usability session notes into a 0–4 severity issue log with x/n counts, task completion with adjusted-Wald intervals, and fixes tied to the metric they should move.

Checks a survey for bias, then reports full distributions with margins of error, NPS with its own error bar, n≥30 segment comparisons, coded open-ends and what not to conclude.

Codes tickets, reviews, NPS verbatims and call notes into a fixed taxonomy, dedupes by account, and returns problems ranked by how many customers have them, with sourced verbatim quotes.

Merges feature requests from sales, CS and calls into underlying jobs, classifies each on evidence, recommends build, workaround, roadmap-candidate or decline, and drafts honest replies.

Tests every stated cancellation reason against usage and billing behavior, codes churn in a fixed taxonomy, and returns preventable patterns with the earliest signal that could have caught each.

Scores a backlog with RICE using anchored Impact, evidence-capped Confidence and sourced Reach, then draws a capacity cut line and tests whether the ranking survives doubt.

Drafts or pressure-tests a two-page product strategy: a graded diagnosis of the crux, a policy that rules options out, coherent actions, a not-do list and kill criteria.

Turns a feature list or backlog into a Now / Next / Later roadmap of problems and outcome metrics with honest commitment labels, plus an external-safe version for customers.

Turns a problem, research notes or tickets into a lean 600–1,000 word PRD where every requirement traces to the problem and every unknown is a labeled slot, not a guess.

Red-teams a PRD or spec on eight anchored dimensions with a quote behind every score, ranks at least five issues by severity, and returns cuts, cross-team questions and a fixed verdict.

Turns a PRD or feature description into INVEST-checked user stories with Given/When/Then criteria, a negative case each, a nine-point edge-case checklist and traceability to requirement IDs.

Tests your north star against delivered customer value, decomposes it into owned drivers with counter-metrics, and reconciles every branch against your own numbers with the math shown.

Designs an A/B test backward from the decision: one primary metric, a sample size that includes power, duration in full weeks, a no-peeking stopping rule and a pre-registered decision table.

Reads pasted A/B test results in a fixed order — SRM check, pre-registration, effect with confidence interval, guardrails, corrections — and ends with one ship, iterate, kill or inconclusive call.

Reads a pasted cohort table, marks immature cells as censored, pools retention by cohort size, finds the churn window, and returns three hypotheses with the test that would confirm each.

Turns tickets and PR descriptions into customer-facing release notes — breaking changes first with migration steps, benefits only where tickets support them, in three paste-ready formats.

Turns raw notes into a bottom-line-first status update with anchored Green/Yellow/Red status, decisions framed as options, ROAM risks, and exec and team versions.

Writes error messages, empty states, confirmations, buttons and onboarding copy that follow fixed patterns, fit every character limit with counts printed, and come with two alternatives each.