wicked-garden

Ask questions about agentic patterns, frameworks, architectures, and best practices

Deep trust and safety audit for agentic systems with risk classification and compliance validation

Interactive agentic architecture design guidance with pattern recommendations and safety validation

Research, compare, and select agentic frameworks with interactive guidance and latest updates

Show available agentic architecture commands and usage

Full agentic codebase review with framework detection, agent topology analysis, and remediation roadmap

Backward compatibility aliases for old multi-plugin command names

Approve a phase and advance to next stage

Archive or unarchive a crew project

Grant or revoke auto-approve (APPROVE-verdict fast-lane) for a crew project

Show artifact convergence lifecycle — states, stalls, and gate verdict

Cut an in-flight crew project over to mode-3 dispatch

Show evidence summary for a task or project

Execute current phase work with adaptive role engagement

Translate jargon-heavy crew output into plain, grade-8 English

Capture user or stakeholder feedback linked to the current crew project

Run QE analysis on a target with configurable rigor

Show available crew workflow commands and usage

Log a production incident linked to the current crew project

Execute remaining work with maximum autonomy and guardrails

Guide for migrating in-flight projects to strict gate enforcement

Enter and manage the operate phase for the current crew project

Configure wicked-crew preferences and working style

Generate a retrospective from operate phase data

Start a new wicked-crew project with outcome clarification

Show current project status, phase, and next steps

Check for quality crisis swarm trigger and recommend coalition response

Alias for crew:auto-approve — grant or revoke APPROVE-verdict auto-advance

Alias for data:analyze — exploratory data analysis and insight generation

Start interactive data analysis session for CSV/Excel files

Data profiling, schema validation, and quality assessment

Show available data engineering commands and usage

ML model review and training pipeline design

Interactive data analysis using DuckDB SQL queries

Sample a dataset and recommend public or custom ontologies based on the data

Data pipeline design and review

Critically analyze any work request before implementation — challenge assumptions, find root causes, identify opportunities, and propose better approaches.

Design statistically rigorous A/B test experiments

Show available delivery management commands and usage

Surface process memory — kaizen status, unresolved action items, aging alerts

Generate multi-perspective delivery reports from project data

Plan progressive feature rollouts with risk assessment

Configure delivery metrics — cost model, commentary sensitivity, and sprint cadence

Add a field to an entity/class and propagate to all affected files

Apply patches from a saved JSON file

Architecture analysis and design recommendations

Systematic debugging session with root cause analysis

Generate or improve documentation for code

Show available engineering commands and usage

Create a new language generator for wicked-patch with scaffolding, tests, and validation

Show what would be affected by a change without generating patches

Review requested changes against codebase and recommend detailed implementation steps

Remove a field and all its usages from the codebase

Rename a field/symbol across all usages in the codebase

Code review with senior engineering perspective on quality, patterns, and maintainability

Show all available wicked-garden domains and commands

Start a new brainstorm session with dynamic focus groups

Structured evaluation tool that uses real external LLM CLIs (Codex, Gemini, OpenCode, Pi) to get genuinely independent model perspectives. Unlike brainstorm (free-form creative exploration), council is a **rigid evaluation tool** for when you have defined options and need a verdict.

Get help with the jam plugin

View a specific persona's contributions across all rounds of a jam session

Get multiple perspectives on a decision without synthesis

Quick exploration with fewer personas and one round

Revisit a past brainstorm decision and record its outcome

View all individual perspectives from a jam session before synthesis

View the full conversation transcript from a jam brainstorm or council session

Manually trigger memory consolidation across tiers

Archive or delete a memory

Show available memory commands and usage

Recall memories matching a query

Backfill search tags on existing memories for better keyword recall

Interactive memory review - browse, understand, and manage stored memories

Show memory statistics

Store a memory for persistence across sessions. Thin wrapper over `wicked-brain-memory` — brain owns the taxonomy, tier derivation, and TTL defaults.

Invoke a named persona to perform a task with a specific perspective

Create or update a custom persona for on-demand invocation

List all available personas with their focus and invocation syntax

Submit a custom persona as a PR to add it to the built-in specialist registry

GitHub Actions workflow generation and optimization

Run contract assertions against plugin subprocess outputs

Audit evidence collection and compliance verification

Regulatory compliance check (SOC2, HIPAA, GDPR, PCI)

Error analysis and pattern detection

GitHub CLI power utilities for workflows, PRs, and releases

System health check and reliability assessment

Show available platform engineering commands and usage

Incident response and triage

Infrastructure review and IaC analysis

View operational logs for the current session

View or set the operational log verbosity level for this session

Run health probes against all installed plugins

Query hook execution traces for the current session

Security review and vulnerability assessment

Discover and query engineer monitoring tools (APM, logging, metrics, cloud)

Distributed tracing analysis for latency and dependencies

Accessibility audit — WCAG 2.1 AA compliance, keyboard navigation, screen reader support, color contrast

Define acceptance criteria from requirements and design

Facilitate stakeholder alignment and consensus building

Analyze customer feedback for themes, sentiment, and trends

Elicit and document requirements through structured discovery

Show available product management commands and usage

Aggregate customer feedback from available sources

Mockup and wireframe generation — ASCII wireframes, HTML/CSS previews, or component specs for developer handoff

Systematic visual design review — design system adherence, spacing, typography, color, component patterns

Screenshot-based UI review using Claude's multimodal vision — analyze layout, color, typography, and consistency from image files

Strategic analysis - ROI, value proposition, market, competitive

Generate actionable recommendations from customer feedback insights

UX and design quality review - flows, UI, accessibility, research

UX flow design and analysis — create user flows, map information architecture, evaluate interaction patterns

File a GitHub issue for a bug, UX friction point, or unmet outcome

Reset wicked-garden to a clean state — choose which data to clear

Analyze dependencies and dependents of a symbol

Show symbol categories — types, layers, directory groupings, and cross-category relationships

Search code symbols only (functions, classes, methods)

Report on lineage coverage and identify symbols without full traceability

Search documents only (PDF, Office docs, markdown)

Show available search and code intelligence commands and usage

Find the most-referenced symbols in the codebase — classes, functions, and modules with the highest connectivity

Analyze what would be affected by changing a symbol (reverse lineage)

Find code that implements a documented feature/section

Build unified index for code and documents in a directory

Trace data lineage from source to sink (UI → DB or reverse)

Run quality crew to validate and improve index accuracy to >=95%

Find where a symbol is referenced and documented

Quick pattern reconnaissance for common code patterns (no index required)

Search across all code and documents

Detect and visualize the service architecture from infrastructure and code patterns

Manage external content sources for the search index

Show index statistics

Validate index accuracy using consistency checks

Configure wicked-garden connection and onboard the current codebase

Session briefing — what happened since last time

Orchestrate multi-AI CLI collaboration — discover, review, council, and persist

Build structured context packages for subagent dispatches

Show session state and recent context for debugging

Import existing domain JSON records into the unified event log

Query the unified event log for cross-domain activity

Show available context assembly commands and usage

Learn a library by fetching docs via Context7 and storing a local cheatsheet

List locally cached library cheatsheets

Intelligent codebase onboarding using the wicked-garden ecosystem

Gather intelligent context from wicked-brain + wicked-garden:search

Five-layer architecture validation, agent topology analysis, orchestration patterns, and framework selection for agentic systems. Focus on structural soundness and scalability. Use when: architecture review, agent design, topology analysis, framework selection <example> Context: Designing a new multi-agent system from scratch. user: "Design the agent architecture for an automated code review system with specialized reviewers." <commentary>Use architect for agentic topology design, scalability review, and structural validation.</commentary> </example>

Framework comparison, selection guidance, migration paths, ecosystem assessment, and latest framework updates. Live research capabilities for emerging frameworks. Use when: framework selection, migration, comparison, latest features <example> Context: Choosing an agentic framework for a new project. user: "Compare LangGraph, CrewAI, and AutoGen for our multi-agent customer support system." <commentary>Use framework-researcher for framework comparison, selection, and migration path analysis.</commentary> </example>

Pattern recognition, anti-pattern detection, refactoring recommendations, design pattern application, and code quality for agentic systems. Use when: code review, refactoring, patterns, anti-patterns, design quality <example> Context: Agentic codebase growing complex and needs design review. user: "Our agent orchestration code is getting tangled. Identify anti-patterns and recommend fixes." <commentary>Use pattern-advisor for agentic anti-pattern detection and design pattern guidance.</commentary> </example>

Token optimization, latency budgets, cost analysis, parallelization opportunities, caching strategies, and context window management for agentic systems. Use when: performance optimization, cost reduction, latency, token usage <example> Context: Agentic system is expensive to run. user: "Our multi-agent pipeline costs $50 per run. Help reduce token usage and cost." <commentary>Use performance-analyst for token optimization, cost reduction, and latency improvements in agent systems.</commentary> </example>

Guardrails, prompt injection defense, PII protection, human-in-the-loop gates, output validation, and hallucination mitigation for agentic systems. Use when: safety review, guardrails, prompt injection, PII, validation <example> Context: New agentic system needs a safety audit. user: "Review our agent pipeline for prompt injection vulnerabilities and missing guardrails." <commentary>Use safety-reviewer for guardrails, prompt injection defense, and PII protection in agent systems.</commentary> </example>

Maintain and strengthen minority positions across sessions for crew projects. Use when: a crew project reaches design phase at complexity >= 4, or whenever the facilitator requests a challenge session. This agent produces and keeps alive the persistent challenge-artifacts.md surface so the dominant direction never crystallises without a steelmanned counter-case. <example> Context: Design phase just produced architecture.md and we are at complexity 5. user: "/wicked-garden:crew:execute" <commentary>Dispatch contrarian to generate phases/design/challenge-artifacts.md with at least 3 themed challenges and a written steelman for each before build is allowed to start.</commentary> </example> <example> Context: challenge-artifacts.md exists but all challenges share the same theme. user: "Continue to build." <commentary>Contrarian detects convergence collapse, adds dissent vectors from other dimensions (security, cost, operability, ethics), and refuses to mark the artifact resolved until variety is restored.</commentary> </example>

Guide outcome clarification through structured inquiry. Use when: defining project outcomes, resolving scope ambiguity before work begins. <example> Context: User wants to add a feature but success criteria are undefined. user: "I want to add real-time notifications to our app." <commentary>Use facilitator to define measurable outcomes and resolve scope ambiguity before design.</commentary> </example>

Archetype-aware phase-boundary evidence evaluator for crew projects. Use when: dispatched at the `testability` or `evidence-quality` gate via gate-policy.json. Reads ctx["archetype"] (from state.extras["archetype"] injected by the dispatcher) and applies the per-archetype score-band table from design.md §1. Emits verdict + score + reason + conditions to gate-result.json and writes one AddendumEntry_1_1_0 record to the reeval addendum. NEVER silent-degrades — missing/invalid archetype triggers a structured warning and explicit code-repo fallback with audit markers. Not for: requirements-quality, design-quality, or any non-target gates.

Fast-path objective gate evaluator for minimal-tier and self-check gates. Use when: a gate-policy.json entry declares `mode: self-check` OR has an empty `reviewers` list OR is `mode: advisory` (findings-only). DO NOT use for full-rigor specialist gates — those dispatch via their declared reviewers (solution-architect, security-engineer, senior-engineer, etc.). <example> Context: A minimal-rigor crew project approves its design phase; gate-policy.json sets `design-quality.minimal` to `mode: self-check` with `reviewers: []`. user: (invoked by phase_manager.approve_phase via _dispatch_gate_reviewer) <commentary> gate-evaluator reads phases/design/design.md, checks byte-count and required-deliverable presence, and emits {verdict: APPROVE|CONDITIONAL, score, reason, conditions}. Never dispatches specialists; never makes subjective calls. </commentary> </example>

Execute implementation tasks according to approved designs and test strategies. Use when: building features, implementing approved designs, tracked development work. <example> Context: Design phase is complete and implementation needs to begin. user: "The design for the caching layer is approved. Start building it." <commentary>Use implementer to execute approved designs with task tracking and quality checks.</commentary> </example>

Independent phase reviewer with cold context. Audits crew phase deliverables, test coverage, and evidence quality — no prior conversation context. Use when: crew phase approval at complexity >= 5, gate review, independent audit of phase artifacts <example> Context: Crew project at complexity 6, design phase awaiting gate approval. user: "Approve the design phase" assistant: "I'll dispatch the independent-reviewer to audit the design artifacts." <commentary>Complexity >=5 triggers cold-context review before phase advancement.</commentary> </example>

Produce the current phase's deliverables and run phase-start / phase-end re-evaluations for a mode-3 wicked-crew project. Use when: phase_manager.execute() dispatches a phase for a full-rigor crew project; the agent receives a phase brief and returns a deliverables manifest plus a parallelization_check block. Re-eval records are written to disk (phases/{phase}/reeval-start.json + reeval-log.jsonl + process-plan.addendum.jsonl). <example> Context: A crew project has just entered the design phase; the clarify gate APPROVED. user: (invoked by phase_manager.execute(project, "design")) <commentary> phase-executor reads the clarify-phase outputs, runs phase-start re-eval (usually a no-op at phase-start), produces phases/design/design.md per the phase template, runs phase-end re-eval appending to reeval-log.jsonl + process-plan.addendum.jsonl, and returns a JSON manifest with files_written, scope_changes, plan_mutations, parallelization_check. </commentary> </example> <example> Context: build phase with 3 independent code-edit sub-tasks. user: (invoked by phase_manager.execute(project, "build")) <commentary> Because phases.json sets phase_executor_may_delegate=true for build, the executor dispatches the 3 edits in a single parallel Task batch (SC-6), aggregates results, writes executor-status.json with sub_agent_timing entries showing overlapping [dispatched_at, completed_at] windows, and returns parallelization_check with dispatched_in_parallel=true. </commentary> </example>

Route to appropriate quality gate. Determines gate type from context, dispatches to gate-specific orchestrators, consolidates results. <example> Context: Project just finished the clarify phase. user: "Run the quality gate — we just finalized our requirements." <commentary>Use qe-orchestrator to detect the appropriate quality gate and dispatch accordingly.</commentary> </example>

Explore codebase and gather context before design or implementation choices. Use when: codebase exploration, pattern discovery, context gathering for decisions. <example> Context: Team needs context on existing patterns before adding a feature. user: "How does our codebase currently handle authentication? We need to add OAuth." <commentary>Use researcher to explore codebase patterns and gather context before design decisions.</commentary> </example>

Perform basic code review and validation. Use when: general code review without a domain-specific specialist available. <example> Context: Implementation is complete and needs a sanity check. user: "Review the changes in the last 3 commits for obvious issues." <commentary>Use reviewer as a fallback for general code review when specialist reviewers aren't matched.</commentary> </example>

Data exploration, statistical analysis, insight generation, and visualization guidance. Helps understand what the data is telling you. Use when: data exploration, statistical analysis, insights, trends <example> Context: Stakeholder wants to understand a metric change. user: "Why did our conversion rate drop 15% last week?" <commentary>Use data-analyst for metric investigation, pattern discovery, and insight generation.</commentary> </example>

Unified data architecture role — owns both operational/transactional data modeling (OLTP schemas, domain entities, data flow, caching, consistency) AND analytical data architecture (warehouse/lakehouse design, star/snowflake/data-vault modeling, governance, partitioning, cost optimization, schema evolution). Combines the former engineering:data-architect + data:analytics-architect. Use when: data modeling (entities + warehouse dimensions), storage selection (OLTP + OLAP), schema design, data flow, data governance, warehouse/lakehouse architecture, partitioning/clustering, cost optimization, schema evolution. <example> Context: Starting a new feature that needs storage AND analytics. user: "Design the data layer for a multi-tenant task system — operational storage and analytics warehouse." <commentary>Use data-architect for the full OLTP + OLAP design in one pass.</commentary> </example> <example> Context: Analytics stack review. user: "Review our data warehouse design — partitioning, cost, governance." <commentary>Use data-architect for warehouse design review, partitioning strategy, and cost optimization.</commentary> </example>

ETL pipeline design, data quality assessment, schema validation, and performance optimization. Reviews data architectures and ensures robust data engineering practices. Use when: ETL pipelines, data quality, schema validation, pipeline optimization <example> Context: Building a data pipeline for a new data source. user: "Design an ETL pipeline to ingest events from our Kafka topics into the data warehouse." <commentary>Use data-engineer for ETL pipeline design, data quality assessment, and schema validation.</commentary> </example>

ML model development, training pipeline design, feature engineering, and deployment guidance. Ensures ML systems are robust, monitored, and maintainable. Use when: ML models, training pipelines, feature engineering, model deployment <example> Context: Building a recommendation system. user: "Design the ML pipeline for our product recommendation engine." <commentary>Use ml-engineer for ML pipeline design, model deployment, and production ML maintenance.</commentary> </example>

Cloud/infrastructure cost analysis AND optimization in one agent. Analyzes billing data to identify cost drivers, anomalies, budget variance, and untagged spend — then recommends right-sizing, reserved capacity, idle-resource cleanup, scheduling, and architectural cost improvements with quantified savings and risk assessment. Use when: cloud cost analysis, billing breakdown, budget variance, cost anomalies, right-sizing, reserved instances, idle resource cleanup, FinOps governance. <example> Context: Monthly cloud bill needs breakdown AND optimization plan. user: "Break down this month's AWS costs and find the top optimization opportunities." <commentary>Use cloud-cost-intelligence for combined cost analysis + optimization with quantified savings.</commentary> </example> <example> Context: Budget variance investigation. user: "We're 30% over budget this quarter — why, and what can we cut?" <commentary>Use cloud-cost-intelligence to identify drivers and produce a prioritized savings plan.</commentary> </example>

Sprint and project delivery management. Track velocity, plan sprints, manage scope, and coordinate cross-team dependencies. Use when: sprint planning, velocity tracking, scope management, project coordination <example> Context: Planning the next sprint. user: "Plan the next sprint based on our velocity and the remaining backlog." <commentary>Use delivery-manager for sprint planning, scope management, and delivery coordination.</commentary> </example>

Design rigorous experiments with statistical analysis. Formulate hypotheses, select metrics, calculate sample sizes, and ensure experimental validity. Use when: A/B tests, experiments, hypothesis, sample size <example> Context: Team wants to validate a new feature with an A/B test. user: "Design an A/B test for our new checkout flow to see if it improves conversion." <commentary>Use experiment-designer for A/B test design, hypothesis formulation, and statistical planning.</commentary> </example>

Track and report progress against milestones, goals, and deadlines. Monitor completion rates, identify slippage, and forecast outcomes. Use when: milestone tracking, progress reporting, deadline monitoring, completion forecast <example> Context: Project milestone approaching and need status check. user: "Are we on track to hit the Q2 milestone? Show me where we stand." <commentary>Use progress-tracker for milestone monitoring, slippage detection, and progress reporting.</commentary> </example>

Risk tracking and escalation management. Identifies delivery risks, tracks mitigation progress, and manages dependency chains. Use when: delivery risks, escalation, dependency tracking <example> Context: Project has multiple dependencies that could cause delays. user: "Identify delivery risks for the platform migration — we depend on 3 other teams." <commentary>Use risk-monitor for delivery risk identification, dependency tracking, and escalation.</commentary> </example>

Coordinate progressive rollouts with risk management. Plan canary deployments, monitor metrics, define rollback criteria, and communicate with stakeholders. Use when: progressive rollout, canary deployment, feature flags <example> Context: New feature needs a safe progressive rollout. user: "Plan a canary rollout for the new search algorithm — start with 5% of traffic." <commentary>Use rollout-manager for progressive rollout planning and go/no-go decisions.</commentary> </example>

Generate multi-perspective stakeholder reports covering Delivery, Engineering, Product, QE, Architecture, and DevSecOps viewpoints. Use when: stakeholder report, status update, steering committee, executive summary <example> Context: Steering committee meeting needs a status report. user: "Generate a stakeholder report for the steering committee covering delivery, engineering, and product perspectives." <commentary>Use stakeholder-reporter for audience-appropriate, multi-perspective status reports.</commentary> </example>

Specialize in API documentation - OpenAPI specs, endpoint documentation, request/response examples, and error documentation. Create comprehensive, accurate API reference materials. Use when: API docs, OpenAPI specs, endpoint documentation, API reference <example> Context: Team is building a new REST API and needs docs for consumers. user: "Generate OpenAPI spec and docs for our user management endpoints." <commentary>Use api-documentarian for API spec generation, endpoint documentation, and reference materials.</commentary> </example>

Backend engineering specialist focusing on APIs, databases, server-side patterns, data modeling, scalability, and integration design. Use when: APIs, databases, server-side code, data modeling, backend architecture <example> Context: Team needs to add a new API endpoint with database persistence. user: "Add a CRUD API for project resources with PostgreSQL storage." <commentary>Use backend-engineer for API + database work, server-side optimization, and service design.</commentary> </example>

Debugging specialist focused on root cause analysis, error investigation, profiling, and systematic debugging strategies. Helps diagnose complex issues. Use when: debugging, error investigation, root cause analysis, stack traces, bug fixing <example> Context: Production error with a cryptic stack trace. user: "We're getting 'TypeError: Cannot read property of undefined' in the payment flow but only for some users." <commentary>Use debugger for root cause analysis, error investigation, and systematic debugging.</commentary> </example>

Developer experience and internal tooling specialist. Owns the quality of the inner dev loop — local environment setup, CI ergonomics, build/test speed, scaffold generators, linter/formatter configuration, IDE integration, and friction removal for day-to-day engineering work. Focus is on multiplying engineer productivity, not on production features. Use when: dev environment setup, local dev tooling, CI speed optimization, build time reduction, test feedback loop, scaffolding generators, linter configuration, pre-commit hooks, friction audit, onboarding time reduction. <example> Context: New engineers take two days to get a working dev environment. user: "Our onboarding is painful. New hires can't run the app on day one." <commentary>Use devex-engineer to audit the inner loop and propose bootstrap script, container, or devcontainer fix.</commentary> </example> <example> Context: CI takes 40 minutes and engineers context-switch constantly. user: "CI is too slow. Cut it to under 10 minutes without losing coverage." <commentary>Use devex-engineer for CI parallelization, caching, test selection, and feedback-loop optimization.</commentary> </example>

Frontend engineering specialist focusing on React, CSS, browser APIs, component design, performance, accessibility, and user experience patterns. Use when: React, CSS, browser APIs, frontend components, UI implementation <example> Context: Building a new interactive dashboard component. user: "Create a filterable data table component with sorting, pagination, and column resizing." <commentary>Use frontend-engineer for React components, UI implementation, and frontend performance.</commentary> </example>

Schema migrations, data backfills, deprecation paths, rollback plans, and zero-downtime cutovers. Specializes in moving a live production system from one shape to another without breaking consumers — dual-write, backfill, cutover, cleanup patterns; expand-contract schema changes; versioned deprecation windows; verifiable rollback plans. Use when: schema migration, data backfill, breaking change rollout, deprecation path, zero-downtime cutover, versioned rollout, rollback planning, database shape change, API version sunset. <example> Context: Splitting a monolith table into two. user: "Plan the migration from orders (single table) to orders + order_items." <commentary>Use migration-engineer for expand-contract plan with dual-write, backfill, cutover, and rollback.</commentary> </example> <example> Context: Sunsetting a deprecated API version. user: "Deprecate /v1/users and move all consumers to /v2/users." <commentary>Use migration-engineer for consumer inventory, deprecation timeline, and cutover plan.</commentary> </example>

Senior engineering perspective on code quality, architecture patterns, maintainability, and implementation guidance. Reviews from developer mindset and provides mentorship on best practices. Use when: code review, refactoring, best practices, implementation guidance, code quality <example> Context: PR is ready for a thorough code review. user: "Review this PR for the new caching layer — focus on correctness and maintainability." <commentary>Use senior-engineer for code review, refactoring guidance, and best-practice mentoring.</commentary> </example>

Design end-to-end solutions with appropriate patterns, technology choices, and trade-offs. Focus on high-level architecture that balances requirements, constraints, and maintainability. Use when: system design, architecture decisions, technology choices, high-level design <example> Context: Greenfield project needs an architecture from requirements. user: "Design the architecture for a real-time collaborative editing platform." <commentary>Use solution-architect for end-to-end architecture design and technology trade-off analysis.</commentary> </example>

Define component boundaries, module organization, and interface contracts. Focus on decomposition and dependency management. Use when: component boundaries, module organization, interface contracts, decomposition <example> Context: Monolithic codebase is becoming hard to maintain. user: "Our 50k-line app has no clear module boundaries. Help us decompose it." <commentary>Use system-designer for component boundary analysis, module decomposition, and interface contracts.</commentary> </example>

Create clear, accessible technical documentation with proper structure, audience awareness, and practical examples. Focus on helping users understand and use the system effectively. Use when: documentation, technical writing, README, user guides <example> Context: New open-source project needs documentation from scratch. user: "Write a getting-started guide for our CLI tool that covers installation, configuration, and first use." <commentary>Use technical-writer for user guides, README files, and operational documentation.</commentary> </example>

Role-plays as focus group personas and synthesizes brainstorming discussions. Use when: brainstorming, ideation, focus group, creative exploration <example> Context: Team needs creative ideas for a new feature. user: "Brainstorm approaches for making our CLI tool more discoverable to new users." <commentary>Use jam facilitator for persona-based brainstorming and structured ideation sessions.</commentary> </example>

Runs structured multi-model council evaluations using external LLM CLIs. Each model responds independently to a fixed question scaffold, then Claude synthesizes. Use when: multi-model evaluation, council vote, independent perspectives, decision validation <example> Context: High-stakes architecture decision needs diverse perspectives. user: "Run a council evaluation on whether to use event sourcing vs. CRUD for our order system." <commentary>Use council for multi-model evaluation and independent perspectives on important decisions.</commentary> </example>

Maintain memory health — decay, archive, and cleanup operations. Use when: memory maintenance, cleanup, decay lifecycle (active -> archived -> decayed -> deleted). <example> Context: User wants to clean up old memories. user: "Clean up old memories" <commentary>Use memory-archivist for decay lifecycle management, cleanup, and memory health checks.</commentary> </example>

Extract and structure learnings from completed tasks. Use when: capturing decisions, patterns, and episodic memories after completing work. <example> Context: Claude just finished a complex debugging session. user: "Great, the bug is fixed!" <commentary>Use memory-learner to capture decisions, patterns, and episodic memories from completed work.</commentary> </example>

Search and retrieve relevant memories without overloading main context. Use when: searching memory store, finding past decisions or patterns, recalling context by topic or tag. <example> Context: User is working on authentication and needs past context. user: "How did we handle JWT validation before?" <commentary>Use memory-recaller to find past implementations, decisions, or patterns by topic.</commentary> </example>

Executes tasks under a named persona's behavioral profile. Receives persona definition (name, focus, personality, constraints, memories, preferences) and a task in the dispatch prompt. Responds from that persona's perspective. Use when: any task needs a specific perspective — review, analysis, advice, brainstorming, content generation. <example> Context: User wants a security-focused review of their auth flow. user: "As the Platform Specialist: review this auth flow." <commentary>Use persona-agent when dispatched by persona:as or --persona flag.</commentary> </example>

Audit evidence collector and control verifier. Gathers artifacts, validates controls, maintains audit trails, and generates compliance reports for certification audits. Use when: audit evidence, compliance artifacts, audit trails <example> Context: Preparing for a SOC2 audit and need to gather evidence. user: "Collect all access control evidence for our SOC2 Type II audit." <commentary>Use auditor for compliance evidence collection, control verification, and audit trails.</commentary> </example>

Failure-mode thinking and resilience testing specialist. Designs chaos experiments, plans game days, injects controlled failures (latency, errors, resource exhaustion, dependency loss), validates graceful degradation, and hardens systems against real incidents before they happen in prod. Use when: chaos engineering, resilience testing, failure injection, game-day planning, fault tolerance review, dependency failure analysis, graceful degradation verification. <example> Context: New service going to prod and leadership wants resilience evidence. user: "Design a chaos experiment for the checkout service before launch." <commentary>Use chaos-engineer to plan a controlled experiment with steady-state hypothesis and blast radius limits.</commentary> </example> <example> Context: Recurring incidents point to fragile dependencies. user: "Plan a game day focused on payments provider outages and database failovers." <commentary>Use chaos-engineer for game-day design, runbook validation, and recovery drill orchestration.</commentary> </example>

Regulatory compliance expert. Evaluates code and systems against SOC2, HIPAA, GDPR, PCI requirements. Identifies violations and provides remediation guidance. Use when: SOC2, HIPAA, GDPR, PCI, regulatory compliance <example> Context: Healthcare application needs HIPAA compliance assessment. user: "Evaluate our patient data handling for HIPAA compliance gaps." <commentary>Use compliance-officer for regulatory assessments (SOC2, HIPAA, GDPR, PCI) and gap analysis.</commentary> </example>

CI/CD pipeline design, workflow automation, and deployment orchestration. Focus on GitHub Actions, GitLab CI, pipeline optimization, and deployment reliability. Use when: CI/CD, pipelines, GitHub Actions, deployment automation <example> Context: Team needs a CI/CD pipeline for a new service. user: "Set up GitHub Actions for our new Python service — lint, test, build, deploy to staging." <commentary>Use devops-engineer for CI/CD pipeline design, optimization, and deployment automation.</commentary> </example>

Incident response specialist focused on rapid triage, root cause correlation, timeline reconstruction, and blast radius assessment during production incidents. Aggregates observability data for fast incident resolution. Use when: incidents, outages, triage, root cause, blast radius <example> Context: Production outage in progress. user: "Users are getting 500 errors on the checkout page. We need to triage immediately." <commentary>Use incident-responder for rapid triage, root cause correlation, and incident analysis.</commentary> </example>

Cloud infrastructure design, Infrastructure-as-Code, scalability, and platform reliability. Focus on AWS/GCP/Azure, Terraform, Kubernetes, and resource optimization. Use when: cloud infrastructure, IaC, Terraform, Kubernetes <example> Context: Deploying a new service to Kubernetes. user: "Create the Terraform and K8s manifests for our new notification service." <commentary>Use infrastructure-engineer for IaC, Kubernetes configuration, and cloud resource management.</commentary> </example>

Discover and validate user needs. Create personas, map journeys, and ensure empathy-driven design. SOLE OWNER of user research activities. Use when: personas, user research, journey mapping, user needs Boundary: user-researcher owns PRIMARY RESEARCH (creating personas, mapping journeys, conducting interviews). For FEEDBACK ANALYSIS on existing data (support tickets, surveys, reviews, NPS), use user-voice instead. <example> Context: Starting a new product and need to understand users. user: "Create user personas for our developer productivity tool." <commentary>Use user-researcher for persona creation, journey mapping, and user needs discovery.</commentary> </example>

Logs, traces, metrics design; SLI/SLO definition; dashboard and alerting architecture. Owns how a system reveals its own state — what's emitted, how it's aggregated, what thresholds fire alerts, and whether the on-call experience is humane. Distinct from SRE (reliability posture) and incident-responder (acute response). Use when: observability design, SLI/SLO definition, logging strategy, distributed tracing design, metrics taxonomy, dashboard architecture, alert-noise reduction, on-call ergonomics, PagerDuty hygiene, OpenTelemetry instrumentation. <example> Context: New service launching and needs observability before production. user: "Design the observability strategy for the recommendations service — logs, metrics, traces, SLOs." <commentary>Use observability-engineer for the full o11y plan: what to emit, dashboards, SLOs, alerts.</commentary> </example> <example> Context: On-call is drowning in noisy alerts. user: "Our on-call got 400 alerts last week and 390 were noise. Fix the alerting." <commentary>Use observability-engineer to audit alert budget and redesign thresholds based on SLOs.</commentary> </example>

Privacy and data protection specialist. Detects PII/PHI, ensures GDPR compliance, implements privacy by design, and protects sensitive data throughout its lifecycle. Use when: PII, PHI, data protection, privacy by design, GDPR <example> Context: New feature collects user data and needs privacy review. user: "We're adding a user profile page that stores name, email, and location. Review for privacy compliance." <commentary>Use privacy-expert for GDPR compliance, PII detection, and privacy-by-design reviews.</commentary> </example>

Release management, versioning strategies, deployment coordination, and rollback procedures. Focus on semantic versioning, changelog generation, deployment strategies, and release orchestration. Use when: releases, versioning, deployment, rollback procedures <example> Context: Preparing a major version release. user: "Prepare the v2.0 release — bump versions, generate changelog, and create the release PR." <commentary>Use release-engineer for version management, changelog generation, and release orchestration.</commentary> </example>

Security scanning and vulnerability assessment from DevSecOps perspective. Focus on OWASP compliance, secure coding practices, secrets management, and defensive security patterns. Use when: security review, vulnerabilities, OWASP, secure coding <example> Context: New API endpoints need a security review before launch. user: "Review the new payment API for OWASP Top 10 vulnerabilities." <commentary>Use security-engineer for vulnerability assessment, secrets scanning, and secure coding review.</commentary> </example>

Site Reliability Engineer focused on system health, capacity planning, performance correlation, and reliability improvement. Use when: reliability, system health, capacity planning, performance analysis <example> Context: Service is approaching capacity limits. user: "Our database is at 80% CPU during peak hours. How should we plan for 2x growth?" <commentary>Use sre for capacity planning, system health assessment, and reliability analysis.</commentary> </example>

Audit accessibility compliance - WCAG guidelines, keyboard navigation, screen readers, semantic HTML, and inclusive design patterns. Use when: accessibility, WCAG, keyboard navigation, screen readers <example> Context: Product needs WCAG 2.1 AA compliance audit. user: "Audit our web app for WCAG 2.1 AA compliance and list the violations." <commentary>Use a11y-expert for WCAG audits, accessible component design, and inclusive patterns.</commentary> </example>

Combined business case + competitive strategy. Builds ROI analysis for technical investments AND analyzes competitive landscape, SWOT, Porter's Five Forces, positioning, and strategic recommendations in one agent. Market sizing inputs (TAM/SAM/SOM) are absorbed here from the former market-analyst. Use when: ROI, business case, investment decision, SWOT, competitive positioning, alternatives analysis, market timing, strategic stance. <example> Context: Team wants to justify a major investment and needs competitive context. user: "Build the business case for our CI/CD product vs. GitHub Actions and CircleCI." <commentary>Use market-strategist to combine ROI + SWOT + positioning in one strategic analysis.</commentary> </example> <example> Context: Product leadership needs strategic stance recommendation. user: "Should we compete head-to-head with Datadog or go niche?" <commentary>Use market-strategist for Five Forces analysis, positioning, and strategic move recommendation.</commentary> </example>

Mockup generation agent. Creates wireframes and design prototypes in ASCII, HTML/CSS, or component spec formats for ideation and developer handoff. Use when: wireframe, mockup, prototype, lo-fi design, component spec, design handoff <example> Context: Quick wireframe needed for a feature discussion. user: "Create an ASCII wireframe for a settings page with sidebar navigation and form sections." <commentary>Use mockup-generator for wireframes, HTML/CSS prototypes, and component specs.</commentary> </example>

Strategic product thinking: roadmap planning, prioritization, trade-offs, and business value alignment. Balances customer needs with delivery capacity. Use when: roadmap, prioritization, product strategy, feature decisions <example> Context: Quarterly planning with too many feature requests. user: "We have 20 feature requests and capacity for 5. Help prioritize." <commentary>Use product-manager for feature prioritization, roadmap planning, and scope definition.</commentary> </example>

Elicit and document requirements with precision. Transform vague ideas into clear user stories with testable acceptance criteria. Use when: user stories, requirements, acceptance criteria, specifications <example> Context: Feature idea needs formal requirements. user: "Write user stories and acceptance criteria for the file sharing feature." <commentary>Use requirements-analyst for user stories, acceptance criteria, and requirements documentation.</commentary> </example>

Visual and UI design review. Evaluates design-system adherence, component patterns, spacing/typography/color correctness, responsive behavior, and visual polish. Reads screenshots directly for rendered-output review; audits code for hardcoded values, inline styles, and token violations. Use when: visual design review, UI consistency audit, design-system compliance, component pattern review, token-migration audit, responsive layout check. <example> Context: New page needs visual consistency review before launch. user: "Review the new pricing page for design system adherence — spacing, typography, and color." <commentary>Use ui-reviewer for design system compliance and token audits.</commentary> </example> <example> Context: Screenshot of a rendered component needs critique. user: "Here's the settings UI screenshot — evaluate visual hierarchy and polish." <commentary>Use ui-reviewer to inspect the rendered PNG directly and score visual quality.</commentary> </example>

User/customer voice specialist. Combines sentiment & theme analysis over raw feedback data with empathy-driven advocacy that translates that signal into product prioritization. One agent for "what are customers saying?" AND "what should we do about it?" Use when: feedback analysis, sentiment & themes, trend detection across support tickets/surveys/reviews, customer empathy, pain-point prioritization, feature prioritization based on customer impact. Boundary: user-voice owns FEEDBACK SIGNAL ANALYSIS (existing data). For PRIMARY RESEARCH (new personas, journey mapping, user interviews), use user-researcher instead. <example> Context: PM wants to understand Q1 feedback themes AND know what to prioritize. user: "What are the main themes in customer feedback this quarter, and which features should we ship next?" <commentary>Use user-voice for combined theme extraction + prioritized recommendations.</commentary> </example> <example> Context: Evaluating churn risk from support data. user: "Are we seeing churn signals in support tickets? Who's affected and how critical?" <commentary>Use user-voice to extract signal, segment impact, and surface retention risks.</commentary> </example>

User experience analysis through research synthesis, journey mapping, usability heuristics, and A/B result interpretation. Distinct from ux-designer (delivery/ execution focus) and ui-reviewer (visual audit focus) — ux-analyst owns the research-to-insight pipeline. Use when: user research synthesis, journey mapping, usability heuristic evaluation, A/B test interpretation, experience gap analysis, UX audit with research grounding. <example> Context: Team has user interview transcripts and wants to understand pain points. user: "Synthesize these 12 user interviews into a UX findings report." <commentary>Use ux-analyst to extract themes, map journeys, and surface usability issues from research.</commentary> </example> <example> Context: A/B test concluded but results need interpretation. user: "We ran an A/B test on the checkout flow. Variant B had +4% conversion but higher drop-off at step 3. What does this mean?" <commentary>Use ux-analyst to interpret the A/B result through a UX lens — behavior signals, journey friction, and next steps.</commentary> </example>

Design and evaluate user flows, interaction patterns, and information architecture. Works generatively (create flows from requirements) and analytically (audit existing flows for usability gaps, Nielsen heuristics, cognitive load, and dead ends). Use when: user flow design, IA mapping, interaction patterns, usability heuristics, flow diagrams (ASCII/Mermaid), navigation hierarchy. <example> Context: New feature needs a user flow designed. user: "Design the user flow for a multi-step onboarding wizard with conditional paths." <commentary>Use ux-designer for user flow generation, IA mapping, and interaction analysis.</commentary> </example> <example> Context: Existing UX needs audit. user: "Review the settings page UX — cognitive load, back navigation, error recovery." <commentary>Use ux-designer to evaluate flows against Nielsen heuristics and surface friction.</commentary> </example>

Value-proposition design AND stakeholder alignment in one agent. Designs value propositions using Jobs-to-be-Done, pain/gain mapping, and differentiation axes; facilitates alignment across stakeholders by surfacing concerns, mediating trade-offs, and building consensus on the value chosen. Use when: value proposition, differentiation, customer benefits, stakeholder alignment, concern surfacing, trade-off mediation, consensus building. <example> Context: New product needs a value proposition AND stakeholder buy-in. user: "Define the value prop for our developer productivity tool and align eng + product on scope." <commentary>Use value-strategist to design the value prop and drive alignment in one pass.</commentary> </example> <example> Context: Teams disagree on the technical approach to deliver value. user: "Engineering wants rebuild, product wants incremental. Help align on what value we're delivering." <commentary>Use value-strategist to mediate trade-offs and build consensus on value.</commentary> </example>

Agentic framework landscape, comparison, and selection guide for choosing the right framework for your use case. Use when: "which agent framework", "compare frameworks", "LangGraph vs CrewAI", "framework selection"

Five-level maturity assessment model for agentic systems from prototype to optimized production. Use when: "agent maturity", "production readiness", "how mature is my agent", "agent assessment"

Systematic review methodology for agentic codebases with issue detection, analysis, and reporting. Use when: "review agentic code", "audit agent system", "check agent quality", "agentic code review"

Trust, safety, and control patterns for production agentic systems with human-in-the-loop gates and guardrails. Use when: "agent safety", "guardrails", "human-in-the-loop", "agent trust", "prompt injection defense"

Adaptive engagement patterns for wicked-crew based on context, phase, and user preferences. Controls autonomy levels (just-finish, balanced, ask-first) and communication style. Use when: "autonomy mode", "crew autonomy", "how much to ask", "just finish it", "ask me before", "adapt to my style", "preference profile", "autonomy level", "just-finish", "ask-first", "balanced autonomy", or setting engagement preferences.

Classifies a list of file paths into change types (ui, api, both, unknown) using an explicit two-pass decision algorithm: extension matching first, then path-segment matching for ambiguous cases. Use when: "detect change type", "classify files", "what kind of change", "ui or api change", "change-type detection", or before creating test tasks.

This skill should be used when running interactive SQL queries against local data files. Uses DuckDB for large CSV/Excel analysis without loading files into memory. Use when: - "query this CSV with SQL" - "run SQL against this data file" - "explore large dataset with DuckDB" - "join these CSV files" - "aggregate across multiple files" - Detecting data quality issues (nulls, duplicates, type mismatches)

This skill should be used when designing or reviewing data pipelines — ETL patterns, orchestration, and performance optimization for data workflows. Use when: - "design a data pipeline" - "review this ETL" - "optimize data processing" - "how should I orchestrate this" - "pipeline architecture"

Design statistically rigorous A/B tests and experiments. Formulate hypotheses, select metrics, calculate sample sizes. Discovers analytics and feature flag tools via capability detection. Use when: "design experiment", "A/B test", "hypothesis", "sample size", "what metrics", "test my feature", "should we experiment"

Guide new developers through team onboarding. Analyzes project health, team patterns, delivery metrics, and codebase structure to produce a personalized onboarding plan with specific first tasks, graduated complexity, and human connections. A guided walkthrough, not an agent identity. Use when: "onboard a new developer", "getting-started guide", "team orientation for new hire", "first week plan", "day-one productivity".

Multi-perspective project delivery reporting with persona-based analysis. Generates actionable reports from project data using 6 specialized stakeholder perspectives. Use when: "generate report", "delivery report", "project status report", "sprint retrospective", "steering committee report", "stakeholder update", "project health summary" Enhanced with: - wicked-garden:mem Stores insights across sessions

Plan and coordinate progressive feature rollouts. Risk assessment, canary deployments, feature flag management, rollback procedures. Discovers deployment tools via capabilities. Use when: "roll out feature", "progressive rollout", "canary deployment", "feature flag", "rollback plan", "launch feature", "deploy gradually"

Define component boundaries, module organization, and interface contracts. Break down systems into cohesive, loosely-coupled components. Use when: "component design", "module boundaries", "how should we organize this", "component interfaces", "dependency management", "system decomposition"

Audit evidence collection and trail verification. Gathers artifacts, validates controls, generates audit reports, and maintains compliance documentation. Use when: "audit trail", "collect evidence", "audit report", "control testing", "compliance documentation"

Compliance analysis for regulatory frameworks (SOC2, HIPAA, GDPR, PCI). Checks code and architecture against compliance requirements, detects violations, and provides remediation guidance. Use when: "compliance check", "SOC2", "HIPAA", "GDPR", "PCI", "regulatory requirements", "is this compliant", "compliance gap"

Error analysis and pattern detection from discovered error tracking sources. Aggregates errors across services, detects patterns, correlates with deployments, and assesses user impact. Use for error investigation and incident response. Use when: "error analysis", "error patterns", "production errors", "error investigation", "why are we seeing errors"

Re-baseline procedure for the AC-11 gate-result benchmark lane (`tests/crew/test_gate_result_benchmark.py`). The benchmark enforces a 2× p95 SLO on `gate-result.json` ingestion. When a deliberate perf change lands on main (validator hardening, cache tuning, schema expansion), the baseline needs updating. Never re-baseline to silence a regression. Use when: "re-baseline AC-11 benchmark", "gate-result benchmark regression", "p95 benchmark baseline out of date", "update benchmark_baseline.json", "benchmark.yml failure", "gate-result p95 exceeds 2x baseline", "rebaseline procedure", or `AC-11` baseline drift.

This skill should be used when working with GitHub CLI (gh) for workflow debugging, PR management, release automation, and repo operations beyond basic git commands. Use when: "gh CLI", "gh run list", "gh pr", "debug CI run", "failed workflow run", "why did CI fail", "PR review queue", "merge PRs", "create release", "repo health check", "bulk issue operations", "workflow runs"

Write secure, optimized GitHub Actions workflows. Security-first approach with performance optimization. Use when: "GitHub Actions", "CI/CD pipeline", "workflow YAML", "deploy pipeline", "automate deployment", "debug workflow", "Actions security"

Write secure, optimized GitLab CI/CD pipelines. Use when: "create CI/CD pipeline", "GitLab CI config", "fix pipeline", ".gitlab-ci.yml", "configure runners", "pipeline optimization"

This skill should be used when working with GitLab CLI (glab) for pipeline debugging, MR management, and release automation. Use when: "glab", "GitLab pipeline", "MR review", "GitLab CI", "merge request", "GitLab release", "glab CLI"

System health overview from discovered observability sources. Aggregates errors, performance metrics, and SLO status across services. Correlates with deployments and code changes. Use for proactive health monitoring and post-deployment validation. Use when: "system health", "health check", "deployment health", "production status", "how is production"

Plugin observability and engineer toolchain discovery. Health probes, contract assertions, and hook execution tracing for the plugin ecosystem. Also discovers and queries APM, logging, metrics, and cloud monitoring CLIs available in the engineer's environment. Use when: "check plugin health", "are hooks working", "silent failures", "trace hook execution", "validate contracts", "plugin diagnostics", "check logs", "query metrics", "view traces", "system monitoring", "datadog", "newrelic", "prometheus", "grafana", "splunk", "cloudwatch"

Policy interpretation and compliance guidance. Translates regulatory requirements into actionable controls, maps policies to code, and provides implementation guidance. Use when: "regulatory policy", "compliance policy", "translate policy to controls", "map regulation to code", "policy implementation guidance"

Detect and transform legacy beta.3 project markers to v6.0 format (D5, AC-13 c). Handles three markers: missing phase_plan_mode, markdown re-eval addendums in process-plan.md (pre-D2 format), and references to the removed legacy gate-bypass env-var in project files. Safe to run on v6-native projects — no markers detected means no-op. Dry-run by default; --apply to write changes. Use when: upgrading a project from wicked-garden v6.0-beta.3 to v6.0; checking whether a project needs migration; transforming legacy artifacts before running crew:approve on a beta project.

Critical thinking framework applied before doing work. Challenges assumptions, reframes problems, identifies hidden opportunities, and validates whether the stated ask is the right ask. A way of approaching work, not a specialist discipline. Use when: "rethink this approach", "is this the right problem", "challenge assumptions", "reframe the problem", "before implementation", "issue triage", "deliberate on this", "should we even do this", "question the approach", "5 whys"

Generate actionable recommendations from customer voice insights. This skill should be used when the user needs to translate analyzed feedback into product priorities, feature recommendations, or strategic guidance. Use when: "recommendations from feedback", "translate feedback to priorities", "feature priorities from customer data", "synthesize feedback into action"

UX flow design and analysis. Creates user flow diagrams, interaction patterns, and information architecture. Generative — creates and maps flows rather than evaluating existing ones (see product/ux-review for evaluation). Use when: "user flow", "flow diagram", "interaction design", "information architecture", "IA mapping", "user journey map", "navigation design", "flow analysis"

UX quality review — user flows, information architecture, user research, and usability evaluation. Works standalone or integrated with wicked-crew. Use when: "UX review", "user flows", "user research", "personas", "user journey", "usability", "information architecture", "navigation review"

Systematic visual design analysis for UI consistency, design system adherence, spacing, typography, color, and component patterns. Use when: "visual checklist", "score the UI", "spacing audit", "typography check", "color palette review", "component pattern audit", "evidence-based design review"

Narrate codebase structure and architecture for orientation — directory layout, key modules, data flows, technical decisions, and code health. A query + synthesis capability rather than a persistent role; produces a guided reading order and flags gotchas for newcomers. Use when: "give me an architecture walkthrough", "narrate this codebase", "explain how this project is organized", "code navigation", "where should I start reading".

Capability router that decides which tools, skills, and agents to use for a task. Discovers CLI tools in PATH alongside MCP servers, skills, and agents. Reduces cognitive load on the main agent by making tool selection decisions. Use when: planning work for unfamiliar domains, evaluating task scope before execution, discovering what integrations or MCP servers could help, CLI detection, which tools are installed, available CLIs, building task execution strategies, or when unsure which tools, skills, or agents to use for current work.

Orchestrates AI-powered brainstorming sessions with dynamic focus groups. This skill should be used when the user wants to brainstorm, explore ideas, get feedback on concepts, or run a focus group discussion. Sessions are tracked as native tasks (process) and stored in wicked-garden:mem (outcome). Use when: "brainstorm this", "explore ideas", "get different perspectives", "focus group", "what do you think about", "pros and cons"

Memory storage, recall, and lifecycle management for maintaining context across sessions. This skill should be used when the user asks to "remember this", "store a decision", "recall what we did", "find past context", "save for later", "what did we learn", "what did we decide", or mentions persisting knowledge, decisions, or learnings. Use when: - "remember this for next time" - "store this decision" - "what did we do before" - "recall past context" - "save this pattern"

Multi-model AI collaboration: discover installed LLM CLIs and orchestrate council sessions, cross-model reviews, and diverse perspective gathering. Detects codex, copilot, gemini, opencode, and pi CLIs at runtime via PATH discovery. Decisions stored in wicked-garden:mem. Transcripts persisted via jam scripts. Use when: - Running multi-model analysis or design review - Getting diverse AI perspectives on a decision - Council sessions with multiple AI models - Second opinion from a different AI - Multi-model code review or architecture critique - "multi-model", "council", "cross-ai", "diverse perspectives", "second opinion"

On-demand persona invocation system for applying named perspectives to any task. Use when: "invoke persona", "act as", "apply perspective", "use a persona", "persona:as", "define persona", "list personas", "custom persona", "review as <role>", "from the perspective of", "--persona flag"

Lead-facilitator rubric. Reads a project description + priors from wicked-brain + the specialist roster, then proposes a full task chain (TaskCreate calls with blockedBy deps and metadata) plus a `process-plan.md` artifact. Replaces the v5 rule engine (smart_decisioning.py + phases.json + SIGNAL_TO_SPECIALISTS) with LLM reasoning over a small number of well-defined factors. Use when: starting a new crew project, re-planning after a gate finding, emitting the initial task chain for `/wicked-garden:crew:start`, or invoked on `TaskCompleted` to prune / augment / re-tier the remaining chain. Also used by `/wicked-garden:crew:just-finish` (yolo mode) to drive autonomous completion.

Five-layer architectural model for building production-grade agentic systems with clear separation of concerns. Use when: "agent architecture layers", "five layer", "agent system design", "separation of concerns agents"

Smart runtime execution for Python and Node scripts with automatic package manager detection. Invoked by other skills and agents when scripts need execution with correct runtime resolution.

On-demand context assembly over wicked-brain + wicked-garden:search. v6 replaced the v5 push-model orchestrator (deleted in #428) with a pull-model skill — subagents call this skill directly when they need a context briefing rather than having one pushed onto every prompt. Use when: "context briefing", "gather background", "what do we know about", "resume where we left off", "catch me up", "what happened before", "context assembly", "prompt enrichment"

This skill should be used when searching code, finding documentation, understanding code-doc relationships, or performing impact analysis. Use when: "search code", "find function", "find class", "where is defined", "search docs", "PDF content", "cross reference", "impact analysis", "blast radius", "what calls this", "find all references". Prefer this over raw Grep/Glob for symbol search, impact analysis, code-doc cross-references, and understanding codebase structure.

This skill should be used when writing, rewriting, or humanizing content. Removes AI tells while preserving meaning. Aligns output to team voice: direct, practical, action-oriented. Use when: "rewrite this", "humanize this", "de-AI this", "make this sound human", "clean up the writing", "too AI-sounding", "write like a human", "draft PR description", "write commit message"

Facilitator-rubric orchestration with hard quality gates.

Core agentic architecture patterns and anti-patterns for building robust multi-agent systems. Use when: "agent pattern", "multi-agent design", "how should I structure agents", "agentic anti-pattern"

Context window management, token optimization, and memory patterns for efficient multi-agent systems. Use when: "context window", "token optimization", "agent memory", "reduce token usage", "context engineering"

Crew-integrated quality gates for phase transitions: value gate, strategy gate, execution gate. Runs gate analysis at crew checkpoints (post-clarify, post-design, post-build) to validate readiness before advancing. Quality checkpoint within the wicked-crew workflow. Use when: "quality gate", "value gate", "strategy gate", "execution gate", "gate analysis", "quality checkpoint", "phase gate", "crew quality gate", "ready to advance", "should we build this", "does it work", "crew QE checkpoint"

Validates that completed task descriptions include required evidence fields at the appropriate level for the task's complexity score. Three tiers (low, medium, high) map to complexity ranges 1-2, 3-4, 5-7. Use when: "validate evidence", "check task completion", "evidence required", "missing evidence", "evidence schema", "task description review", or validating a TaskUpdate description before marking complete.

Translate jargon-heavy crew output into plain language. Input is typically a gate finding, reviewer brief, phase summary, or process plan containing specialist vocab (RED, BLOCK, convergence, blast radius, parallelization_check, CONDITIONAL, BLEND rule, etc.). Output is 2-4 sentences at a grade-8 reading level with no specialist vocab left behind. Use when: "explain this", "in plain English", "what does this mean", "translate for me", "simplify", "dumb this down", or any request to render crew jargon into language a non-practitioner can act on. Also used automatically by the orchestrator when `crew.output_style = paired` or `plain-only` — the skill produces the `**Plain:**` line.

Automated GitHub issue detection and filing from Claude sessions. Tracks tool failures and task completion mismatches. Files issues automatically at session end or on demand via /wicked-garden:report-issue. Includes duplicate detection, codebase research, memory recall, SMART criteria validation, and advisory quality gate before filing. Use when: "file a bug", "report issue", "something went wrong", "not working as expected", "create issue", reporting UX friction, logging unmet outcomes, or investigating tool failures.

Generates test task creation parameters from a detected change type and implementation task subject. Routes to the correct test type (visual/endpoint) and produces TaskCreate-ready subjects, descriptions, and metadata. Use when: creating test tasks after change-type detection, QE task generation, "test task factory", "create test tasks", "generate QE tasks", or after change-type-detector classifies files as ui/api/both.

This skill should be used when exploring datasets for patterns and generating business insights — exploratory analysis, statistical methods, and visualization guidance. Use when: - "analyze this data" - "what patterns are in the data" - "generate insights from" - "exploratory data analysis" - "what does this data tell us"

This skill should be used when profiling datasets, validating schemas, or assessing data quality. Covers dataset profiling, schema validation, and quality assessment. Use when: - "profile this dataset" - "validate schema" - "check data quality" - "what's in this CSV/Excel file" - "describe this data"

This skill should be used when working with machine learning models — architecture review, training pipeline design, feature engineering, and deployment guidance. Use when: - "review this ML model" - "design ML training pipeline" - "how should I deploy this model" - "feature engineering advice" - "ML architecture guidance"

Complete solution architecture design with patterns, decisions, and diagrams. Define overall system structure, technology choices, and architectural trade-offs. Use when: "design the architecture", "what's the overall structure", "architecture patterns", "technology stack", "system architecture"

Backend engineering specialist for APIs, databases, server-side patterns, data modeling, scalability, and integrations. Use when: "design an API", "database schema", "how to handle auth", "backend performance", "REST endpoint", "query optimization", "server-side architecture"

Systematic debugging and root cause analysis for investigating errors, diagnosing complex issues, and performance profiling. Use when: "debug this error", "why is this failing", "root cause analysis", "fix this bug", "investigate crash", "stack trace", "not working"

Audit documentation coverage and quality. Find undocumented code, assess completeness, and provide coverage metrics. Use when: "audit docs", "doc coverage report", "find undocumented code", "documentation gaps", "what's missing docs", "doc completeness"

Senior engineering guidance on code quality, architecture patterns, and best practices. Use for implementation planning, code review, or general engineering questions about maintainability and design. Use when: "review this code", "how should I implement", "is this good practice", "code quality", "refactor", "clean up", "best practice", "design pattern"

Frontend engineering specialist for React, CSS, browser APIs, component design, accessibility, and performance. This skill should be used when the user asks about UI implementation, component review, or frontend-specific questions. Use when: "build a React component", "CSS layout issue", "frontend performance", "browser API", "component design", "accessibility check", "responsive design"

Generate documentation from code - extract types, comments, and signatures to create API docs, README files, and reference documentation. Focus on useful, actionable docs. Use when: "generate docs", "create documentation", "document the API", "generate README", "make docs from code"

Design API contracts, service boundaries, and integration patterns. Define how components and services communicate and interoperate. Use when: "API design", "service integration", "how do these communicate", "API contract", "integration pattern", "REST API", "GraphQL", "event schema"

Language-agnostic code generation and change propagation. Use this skill when adding fields to entities, renaming symbols across files, or propagating changes that affect multiple files. Use when: "add field", "rename everywhere", "propagate change", "generate migration", "update all references". This is the CODE MUTATION counterpart to wicked-garden:search (which is read-only).

Keep documentation in sync with code changes. Detect stale docs, identify drift between code and docs, and suggest updates when code changes. Use when: "check doc freshness", "sync docs", "docs out of date", "update docs after refactor", "docs drift", "docs out of sync"

Diagnose missing tools and dependencies, offer to install them. Use when: "command not found", "ModuleNotFoundError", "missing tool", "install dependency", "prereq check", "setup validation"

Distributed tracing analysis for performance investigation and service dependency mapping. Analyze latency breakdowns, identify bottlenecks, map service calls, and correlate slow requests with code changes. Use when: "trace analysis", "distributed tracing", "latency investigation", "slow requests", "service dependencies", "performance tracing"

Define testable acceptance criteria from requirements and design. Bridge product requirements with QE test scenarios. Use when: "define acceptance criteria", "how do we know it's done", "what should QE test", "definition of done"

Accessibility audit, compliance evaluation, and implementation guidance — WCAG 2.1 AA, keyboard navigation, screen readers, ARIA patterns, and color contrast. Use when: "accessibility audit", "a11y compliance", "WCAG evaluation", "audit for accessibility", "remediation plan", "compliance report", "make this accessible", "keyboard navigation", "screen reader support", "color contrast", "ARIA patterns", "semantic HTML", "accessible component"

Sentiment analysis, theme extraction, and trend detection from customer feedback. Use when: "what are customers saying about X", "feedback trends", "analyze customer feedback", "sentiment analysis", "customer themes", "what's the top complaint"

Visual design and UI consistency review — design system adherence, spacing tokens, typography, color usage, component patterns, and responsive polish. Use when: "design critique", "polish the UI", "design system adherence", "responsive design", "visual polish", "review the design", "UI inconsistency"

Strategic business analysis for technical investments. Evaluates ROI, value proposition, competitive positioning, and market alignment. Provides decision support with business justification. Use when: "competitive analysis", "market strategy", "ROI", "business case", "value proposition", "strategic alignment", "investment decision"

Visual asset lifecycle management — analysis, generation, modification, and review. Three sub-skills cover the full creative pipeline with provider abstraction. Use when: "image", "visual", "generate", "review image", "edit image", "create image", "analyze image", "brand check", "accessibility review"

Aggregate customer feedback from discovered sources across support, surveys, social, and direct channels. This skill should be used when the user needs to gather customer voice data to inform product decisions or understand sentiment. Use when: "listen to customers", "what are customers saying", "gather feedback", "customer sentiment", "aggregate feedback", "voice of customer"

Digital mockup and wireframe generation. Outputs ASCII wireframes for quick ideation, HTML/CSS previews for interactive review, or component specs for developer handoff. Use when: "mockup", "wireframe", "prototype", "lo-fi design", "component spec", "design sketch", "UI layout", "layout draft", "design handoff"

Strategic product thinking for roadmap, priorities, and business value. Elicit requirements, define scope, and align stakeholders. Works standalone or integrated with wicked-crew. Use when: "product strategy", "roadmap", "product backlog", "what should we build next", "scope definition", "feature prioritization"

Deep requirements elicitation and user story definition. Transform vague ideas into clear, testable specifications. Use when: "write user stories", "define requirements", "elicit requirements", "clarify scope", "specification document"

Graph-based requirements as atomic markdown files with rich frontmatter. Filesystem-as-graph: each AC is a node, directories are groupings, meta.md files are the human interface, frontmatter traces are the edges. Use when: "requirements", "user stories", "acceptance criteria", "elicit requirements", "define requirements", "graph requirements"

Convert monolithic requirements documents into graph-structured requirements directories. Parses user stories and acceptance criteria from existing docs and creates atomic graph nodes. Use when: "migrate requirements", "convert to graph", "split requirements", "restructure requirements"

Navigate, query, and maintain requirements graphs. Regenerates meta.md on demand, checks coverage, finds gaps, and lints the graph structure. Use when: "show requirements", "coverage report", "requirements status", "refresh meta", "lint requirements", "find gaps", "navigate requirements"

Screenshot-based UI review using Claude's multimodal capabilities. Analyzes visual design from image files — layout, spacing, color, typography, responsiveness — and compares against design system rules. Use when: "screenshot review", "UI screenshot", "visual analysis from image", "design from screenshot", "review PNG", "review JPG", "image-based review"

21 hooks across 12 events