Youdede toolkit

Plugin youdede-toolkit in youdede/toolkit. Commands read from the repo's manifest — today.

Skills (169)

• assess-learning-gaps

  Assess a learner with adaptive questions tied to their mission, record evidence and calibration, and update a focused learning plan. Use when the user asks to gauge knowledge or identify what to learn next.

• assess-mutation-test-value

  Plan or assess a bounded mutation-testing exercise by selecting consequential code, defining budgets, triaging survivors, and improving observable assertions. Use when coverage is high but test effectiveness remains uncertain.

• assess-story-narrative-provenance

  Measure whether a full English short story, from a file or a public web page, reads as human-written or AI-generated at the level of narrative choices, using a 304-feature narrative annotation and a classifier trained on human and AI fiction, and report the probability with its uncertainty and valid

• audit-agent-guidance-drift

  Audit AGENTS.md, CLAUDE.md, and equivalent scoped guidance for stale paths, false commands, contradictory precedence, or newly durable invariants. Use before integration or after structural repository changes when agent instructions may have drifted.

• audit-agent-workflow-opportunities

  Analyze one agent run or a consented bounded window of local agent history to find repeated corrections, manual sequences, and preventable failures worth encoding. Use when deciding what should become guidance, a skill, hook, subagent, tool, or automation.

• audit-ai-search-visibility

  Audit a public site for AI-answer discoverability using crawl access, entity clarity, source quality, extractable answers, citation evidence, and measurement limits. Use when a brand wants to understand or improve visibility in AI-generated search without relying on speculative ranking tricks.

• audit-app-store-listing

  Audit an Apple App Store or Google Play listing using observable metadata, visual assets, review themes, category context, localization, and conversion hypotheses. Use before changing a listing or planning an app-store experiment.

• audit-design-system-consistency

  Audit specified UI surfaces for coherent token, component, state, responsive, and accessible-semantic use with evidence and explicit uncertainty. Use when a design system or component set needs consistency review.

• audit-ethical-persuasion

  Audit marketing copy and interaction design for persuasive mechanisms, user comprehension, claim support, autonomy, accessibility, and dark-pattern risk. Use before shipping a conversion experience or when trust and pressure tactics need independent review.

• audit-implicit-decisions

  Audit the current work for consequential choices that were made implicitly or with low confidence, then present alternatives and evidence. Use before approval, handoff, or when the user asks what assumptions remain.

• audit-knowledge-state-freshness

  Audit a bounded notes collection, project memory, or always-loaded context for current-state claims contradicted by newer evidence. Use when an agent gives stale answers or a knowledge base may mix mutable state with append-only events.

• audit-published-content-defects

  Find the defects a reader can see on published web pages, such as markdown left visible, unresolved placeholders, double-escaped entities, paragraphs cut after an abbreviation, repeated paragraphs, and passages left untranslated, across one page, a whole sitemap, or a folder of saved HTML, and repor

• audit-safe-browsing-readiness

  Assess whether a public website is ready for reputation and malware-review systems using observable security, ownership, content, and indexing signals. Use before submitting a site for review or diagnosing a warning.

• audit-technical-seo

  Audit a website for technical and on-page search issues across crawl access, canonicalization, rendering, index signals, templates, internal links, content, structured data, and performance. Use when visibility declines or before a substantial SEO change.

• author-deterministic-agent-hook

  Design, implement, and validate a deterministic coding-agent hook from a precise invariant or observation requirement. Use when a repeated must-block or must-observe boundary should be enforced through a runtime-native hook rather than prompt compliance.

• author-effective-agent-skill

  Create or improve a portable Agent Skill with a narrow trigger, lean procedure, explicit boundaries, progressive disclosure, deterministic helpers, and forward validation. Use for SKILL.md authoring and review.

• author-guided-setup-wizard

  Author a bounded shell wizard for manual setup steps that require human interaction, credentials, dashboards, or irreversible confirmation. Use when automation cannot complete the procedure but a repeatable guided path is valuable.

• author-mcp-server-contract

  Design or review an MCP server’s resources, prompts, tools, schemas, transport, authentication, pagination, errors, and authority boundaries before implementation. Use when exposing a real system to coding agents through the Model Context Protocol.

• build-competitor-evidence-profile

  Build structured competitor evidence profiles from lawful public sources, supplied materials, and reproducible observations. Use when positioning, product, sales, or strategy work needs comparable current facts rather than an untraceable competitive narrative.

• build-compliant-prospect-list

  Build or review a B2B prospect list using a defined ideal-customer situation, lawful public business sources, explicit qualification, provenance, minimization, and suppression. Use before approved outbound work when target accounts require evidence rather than bulk contact collection.

• build-disposable-prototype

  Build a disposable logic or interface prototype around one explicit uncertainty, with observable state, representative scenarios, and a removal decision. Use when cheap interaction evidence is more useful than further abstract discussion.

• build-macos-metrics-collector

  Design and validate a native macOS metrics collector using supported system interfaces, bounded sampling, SQLite, and launchd. Use for local device observability where privacy and low overhead are required.

• build-marketing-operating-plan

  Build a marketing operating plan from business stage, customer evidence, growth constraint, economics, positioning, capacity, and measurement. Use when a team needs coordinated priorities and operating decisions rather than a broad collection of tactics.

• build-openapi-contract

  Build or repair an OpenAPI document by reconciling implementation, tests, product requirements, authentication, schemas, examples, and error behavior. Use when an HTTP API needs an accurate machine-readable contract or drift review.

• build-progressive-learning-exercises

  Build a progression of learning exercises from objectives and prerequisite skills, with learner prompts, reference solutions, feedback, accessibility, and structural validation. Use when authoring workshops, courses, or practice repositories.

• build-repository-quality-gates

  Build or repair repository quality gates by discovering existing tools, classifying checks by latency and risk, and proving clean and failing paths. Use when commits or pull requests need reliable automated feedback without framework lock-in.

• build-sales-enablement-asset

  Build or review a sales deck, one-pager, battle card, talk track, demo brief, objection guide, or ROI worksheet for a defined buyer and deal moment. Use when sellers need an evidence-backed asset they can apply consistently.

• capture-content-idea

  Capture one or more content ideas into a configured append-only backlog, preserving the user’s wording and recording source context. Use for video, article, podcast, or campaign ideas, not general tasks.

• compose-visual-layout

  Compose the spatial structure of a page or screen, covering density, grid, spacing rhythm, alignment, grouping, hierarchy, emphasis for the primary action, and layered depth, with every choice justified by the content's priority order. Use when arranging a new screen or rebuilding an existing layout

• compress-response-with-context

  Rewrite the previous answer into a shorter plain-language version with an optional compact context recap, preserving decisions, caveats, evidence, and next action. Use when the user asks for shorter or simpler.

• conduct-source-backed-research

  Conduct deep source-backed research using available browsing or research tools, then synthesize a cited report that separates facts, inference, and uncertainty. Use for consequential questions that need current external evidence.

• critique-ui-surface

  Critique a rendered screen or mockup against explicit criteria, covering hierarchy, composition, typography, color use, information density, affordance, and brand consistency, and return a prioritized fix list with anchored severity, effort estimates, and evidence for each finding. Use when a visual

• debug-ci-failure

  Diagnose a failing CI job by correlating an exact run, logs, workflow revision, environment, and repository evidence before proposing a bounded remedy. Use when a check fails remotely, flakes, or differs from a local result.

• debug-systematically

  Diagnose bugs, test failures, build failures, performance regressions, and unexpected behavior through reproducible evidence and root-cause tracing before proposing a fix. Use when a failure is unclear, a previous fix failed, or multiple components may be involved.

• decompose-dependent-work-slices

  Decompose a plan or specification into bounded vertical work items, explicit acceptance evidence, and verified blocking edges. Use when work must be delegated or sequenced without losing end-to-end value.

• define-domain-language

  Define or repair shared domain language by reconciling stakeholder terms, code behavior, concrete scenarios, and durable decisions. Use when overloaded vocabulary or undocumented trade-offs make design and implementation ambiguous.

• define-revenue-lifecycle

  Define or audit the lead-to-revenue lifecycle across marketing, sales, success, billing, and analytics using explicit stage semantics, ownership, service levels, routing, data quality, and reconciliation. Use when handoffs or pipeline reporting are inconsistent.

• delegate-bounded-subtasks

  Delegate bounded independent subtasks to available subagents while preventing file overlap, missing context, and unverified integration. Use when parallelism materially reduces latency and each task has a clear output.

• delegate-to-codex-cli

  Delegate an independent coding or review task to Codex CLI with a narrow brief, isolated worktree, bounded permissions, and parent-agent verification. Use when the CLI is available and parallel delegation adds value.

• deliver-incremental-change

  Break and deliver a bounded implementation in vertical, behaviorally verifiable slices with explicit invariants, checkpoints, and integration evidence. Use when a change is too large for one safe edit but can progress without speculative scaffolding.

• deliver-test-first

  Implement features, bug fixes, refactors, and behavior changes through a verified red-green-refactor loop using the repository's own test stack. Use when code behavior will change and regression evidence is required before completion.

• design-agent-evaluation-harness

  Design or audit an evaluation harness for a coding or tool-using agent using isolated tasks, controlled environments, deterministic checks, calibrated judgment, traces, and regression thresholds. Use before changing agent prompts, tools, policies, or orchestration.

• design-churn-interventions

  Design retention interventions across cancellation, failed-payment recovery, pause, downgrade, support, and win-back paths. Use when a subscription business needs to diagnose churn mechanisms and test remedies without obstructing cancellation.

• design-community-growth-system

  Design or audit an online customer, practitioner, or product community around member value, governance, programming, moderation, and sustainable growth. Use when launching or repairing a forum, Slack, Discord, ambassador group, or community-led support program.

• design-controlled-experiment

  Design a controlled product or marketing experiment with a falsifiable hypothesis, primary metric, guardrails, exposure plan, and decision rule. Use when a team is comparing variants or needs a defensible experiment rather than an informal before-and-after check.

• design-data-pipeline-change

  Design or review a batch, streaming, ETL, ELT, or dbt pipeline change across producer and consumer contracts, idempotency, backfills, dual running, reconciliation, cutover, and recovery. Use before changing a governed data flow or replaying historical data.

• design-data-quality-controls

  Design or audit data-quality controls for datasets, pipelines, metrics, or interfaces using semantic contracts, risk, checks, lineage, ownership, observability, and incident response. Do not use it to implement pipelines, choose vendors, or certify compliance. Use when downstream decisions require t

• design-deep-module-interface

  Design or improve a module interface by measuring caller knowledge, variation, failure semantics, locality, and test leverage. Use when a module is shallow, cross-file changes scatter, or the correct test seam is unclear.

• design-delegable-workflow

  Design a recurring workflow for human, deterministic automation, and optional agent work by modeling triggers, state, decisions, authority, checkpoints, failures, and observability. Use before automating a repeated operational loop.

• design-evidence-backed-offer

  Design or review a commercial offer by aligning a real customer outcome with scope, delivery mechanism, proof, price structure, risk allocation, capacity, and clear terms. Use when packaging a service, course, productized engagement, or high-consideration sale.

• design-for-localization

  Specify how an interface survives locale change through banded text growth allowances, string-structure constraints, bidirectional mirroring rules with mixed-direction handling, locale-driven date, number, name, and address formats, and review of icon and color meaning across cultures, each stated a

• design-form-experience

  Design a form that people finish, covering field necessity, grouping and sequencing, input affordances, validation timing, error recovery, and completion confidence. Use when creating or revising any data-collection flow, from a single input to a multi-step wizard.

• design-interface-feedback

  Design how an interface acknowledges people, with response timing budgets, waiting and progress patterns, success and failure signals, empty states, interrupted-task recovery, and error messages that name the problem and the next step. Use when specifying loading, confirmation, notification, or erro

• design-lifecycle-email-sequence

  Design or review welcome, activation, nurture, retention, transactional, win-back, or re-engagement email sequences using explicit states, value, consent, and suppression logic. Use when a lifecycle program needs more than isolated email copy.

• design-llm-evaluation

  Design or audit an evaluation for an LLM feature, agent, retrieval flow, or prompt using representative tasks, reference evidence, deterministic checks, calibrated judges, uncertainty, and release decisions. Do not use it to tune prompts or certify a model, and do not run model calls without separat

• design-marketing-measurement

  Design or audit marketing measurement with decision questions, event semantics, identity boundaries, consent constraints, QA, and ownership. Use when teams need trustworthy conversion tracking or a tracking plan rather than an indiscriminate list of events.

• design-measurable-growth-loop

  Design or audit a product, content, referral, collaboration, or data growth loop using explicit actors, state transitions, incentives, delays, leakage, and measurement. Use when a team believes an action can create the next qualified input and needs to test whether it truly compounds.

• design-navigation-and-search

  Specify how users move through a product and retrieve its content, covering navigation pattern selection, menu ordering, wayfinding state, and search behavior from query entry to empty results. Use when a product needs its navigation chosen, its menus ordered, or its search experience specified.

• design-observability-signals

  Design or audit observability for a service, job, client, or distributed flow using operational questions, failure modes, structured events, metrics, traces, alerts, retention, and privacy. Use before implementation or when operators cannot explain system behavior safely.

• design-onboarding-journey

  Design a first-run experience that carries a new user from arrival to the first moment of real value, with minimum required setup, safe defaults, progressive disclosure, honest progress signals, and a measurable activation definition. Use when planning onboarding flows, setup wizards, new-account em

• design-pricing-and-packaging

  Design or review price levels, value metrics, packages, trials, freemium, discounts, and migration using customer evidence, unit economics, segmentation, and operational constraints. Use when deciding what to charge or how plans should differ.

• design-private-data-lifecycle

  Design or review a privacy-aware data lifecycle across collection, classification, processing, access, sharing, retention, deletion, backup, and incident handling. Do not use it to give legal advice, certify compliance, or approve a policy exception. Use when a product or pipeline handles personal,

• design-property-and-fuzz-tests

  Design property-based or fuzz testing from observable invariants, input models, shrinking, reproducible seeds, budgets, and failure triage. Use when example tests miss broad input spaces, parser boundaries, state transitions, or adversarial inputs.

• design-qualified-lead-magnet

  Design a guide, template, benchmark, checklist, diagnostic, or resource that creates standalone value and attracts qualified demand. Use when a team needs a lead magnet aligned with a real audience job and a consent-aware follow-up path.

• design-referral-program

  Design or audit a customer referral, affiliate, or ambassador program using product fit, participant value, consent, incentive economics, attribution, disclosure, fraud controls, and operational ownership. Use before launching or materially changing a referral system.

• design-respectful-conversion-overlay

  Design or review a popup, modal, banner, slide-in, or sticky conversion prompt with contextual value, accessible behavior, clear dismissal, frequency limits, and consent-aware measurement. Use when an interruptive surface is being considered or underperforming.

• design-responsive-behavior

  Specify how one interface serves every viewport through content-driven breakpoints measured in the right frame (viewport or container), reflow rules that preserve priority and meaning, concrete touch target minimums and input adaptation, and platform divergences, each written as testable behavior pe

• design-safe-programmatic-seo

  Design or audit a programmatic SEO system using reliable data, a real repeated user job, differentiated page value, template quality, indexing controls, and lifecycle ownership. Use before generating many search landing pages.

• design-service-level-objectives

  Design or audit service-level indicators, objectives, error budgets, windows, alerts, and operating policy around critical user journeys. Use when a team needs a reliability target that can guide engineering and incident decisions.

• design-stable-api-contract

  Design or review an internal, public, HTTP, event, library, or command interface using consumer evidence, invariants, compatibility, error semantics, and evolution rules. Use before implementation or when an interface change could affect independent consumers.

• design-stakeholder-questionnaire

  Design a stakeholder questionnaire that closes a defined knowledge gap with prioritized, single-purpose questions and usable answer fields. Use when the user must obtain facts or decisions from someone with unique context.

• document-brand-voice

  Create or update a reusable brand-voice guide from approved examples, audience evidence, strategic context, linguistic patterns, and explicit boundaries. Use when multiple writers or agents need consistent choices without flattening all content into one generic tone.

• document-product-positioning

  Create or update a project-local product positioning record covering audience, situation, alternatives, differentiated value, proof, language, constraints, and open questions. Use before substantial marketing work when foundational context is missing or inconsistent.

• document-project-vision

  Interview a project owner and write concise vision documentation plus ADRs for durable decisions. Use when essential intent lives in a person’s head and future contributors need trustworthy context.

• edit-ai-draft-for-human-voice

  Revise an AI-assisted text, from a single paragraph to a full story, essay, email, or post, in English or Italian, so it reads less like default model prose without adding any fact, changing structural choices first and surface style second, then verify the change with a bundled signal measurement a

• edit-marketing-copy

  Edit existing marketing copy for clarity, specificity, evidence, voice, structure, risk, and action while preserving approved meaning. Use when copy already exists and needs a reviewable improvement rather than a new strategy or unsupported rewrite.

• enforce-typescript-module-boundaries

  Inspect and enforce TypeScript module boundaries so consumers import through declared entry points while internals remain local. Use when deep imports, package cycles, or test access make contracts and refactors unreliable.

• establish-color-system

  Define a product color system, including tonal scales built in a perceptual color space, semantic roles, state colors, dark scheme derivation with overlays and scrims, and multi-theme constraints, with contrast computed per pairing at authoring time and certification routed to accessibility. Use whe

• establish-design-tokens

  Establish the token layer of a design system, covering tier architecture, naming grammar, semantic role coverage, contrast-paired roles, theme and mode support, token format and build pipeline, and a migration path away from hardcoded values. Use when creating or restructuring design tokens, or befo

• establish-typography-system

  Define a typographic system, including a type scale with per-range ratios or hand-tuned steps, role assignments, line height and measure rules for reading comfort, fluid responsive type behavior, and typeface pairing constraints, expressed as reusable rules rather than per-screen choices. Use when s

• evaluate-agent-guidance-value
• evaluate-code-review-feedback
• evaluate-free-tool-growth
• evaluate-retrieval-system
• execute-verifiable-plan
• extract-youtube-captions
• frame-design-direction
• generate-marketing-experiments
• govern-design-system-evolution
• guide-stepwise-setup
• implement-valid-structured-data
• improve-product-activation
• interview-for-priorities
• interview-to-build-brief
• isolate-work-with-worktrees
• keep-macos-awake
• load-architecture-decisions
• make-practical-decision
• map-conversion-site-architecture
• map-experience-journeys
• map-multi-session-decisions
• optimize-with-performance-profile
• plan-accessibility-conformance-evidence
• plan-channel-native-social-content
• plan-compatible-deprecation
• plan-dependency-upgrade
• plan-earned-media-outreach
• plan-evidence-based-design-research
• plan-evidence-led-content
• plan-infrastructure-change
• plan-load-test
• plan-paid-acquisition
• plan-product-launch
• prepare-branch-integration
• prepare-eaa-conformance-evidence
• prepare-reviewable-pull-request
• produce-testable-ad-creative
• provision-postgres-readonly-role
• publish-agent-skill
• qualify-co-marketing-partner
• qualify-influencer-partnership
• query-production-readonly
• reconcile-marketing-attribution
• reduce-signup-friction
• request-independent-code-review
• research-online-purchase
• resolve-merge-conflicts-by-intent
• resolve-next-decision
• review-agent-tool-boundaries
• review-auth-boundary
• review-ci-workflow-change
• review-conversion-friction
• review-database-migration-plan
• review-frontend-quality
• review-plan-execution-divergence
• review-schema-evolution
• review-supply-chain-change
• review-upgrade-moment
• schedule-agent-runs
• simplify-code-with-invariants
• specify-component-contracts
• specify-engineering-handoff
• structure-information-architecture
• surface-build-decisions
• sync-skills-across-agents
• synthesize-customer-language
• synthesize-user-research-evidence
• teach-with-retrieval-practice
• test-interface-in-browser
• test-screen-reader-journey
• triage-engineering-request
• verify-cognitive-access-barriers
• verify-completion-evidence
• verify-mobile-release-readiness
• verify-production-release
• write-article-from-source-material
• write-compliant-cold-email
• write-conversion-copy
• write-folder-agent-guidance
• write-honest-comparison-page
• write-implementation-plan
• write-incident-runbook
• write-learning-postmortem
• write-project-agent-guidance
• write-release-notes-from-evidence
• write-session-handoff
• write-source-backed-research-brief
• write-verifiable-goal-loop
• write-verifiable-specification

Similar plugins