Skill Catalog

All slash commands across the four plugins -- 76 skills total

Start Here

7 Skills That Cover 90% of Use Cases

New to KAI? Start with these. Everything else is for fine-tuning specific phases.

Pick Your Entry Point

You don’t need to memorize 76 skills. These 7 cover the full development lifecycle. Each one is a coordinator that handles its phase end-to-end, including sub-steps.

Skill	What It Does	When to Use
/dx-req <id>	Fetch ticket, validate readiness, distill requirements, research codebase	Starting any story or feature
/dx-plan	Generate implementation plan with step tracking	After requirements, before coding
/dx-step-all	Execute all plan steps with auto-fix on failure	Hands-free implementation
/dx-pr	Commit, push, create PR linked to ticket	Code is done, ready to submit
/dx-agent-all <id>	Full pipeline: requirements through PR in one command	Simple stories, hands-free delivery
/dx-figma-all <url>	Extract design, generate prototype, verify visually	Figma design-to-code
/dx-bug-all <id>	Triage, fix, verify, PR — full bug lifecycle	Bug reports

Partial Flow? Jump In Anywhere

Each skill reads from local spec files, not from the previous skill’s session. So you can:

Already have requirements? Skip to /dx-plan
Already have a plan? Skip to /dx-step-all
Coded manually? Skip to /dx-step-verify then /dx-pr
Just need a PR? Run /dx-pr directly

Overview

Full Skill Inventory

All skills grouped by plugin and category. The 7 primary skills above are marked with a star.

dx-core

Platform-agnostic ADO/Jira workflow: requirements, planning, execution, review, PR, bug fix, documentation, and utility.

49 skills

dx-aem

AEM full-flow: component dialogs, JCR content, editorial QA, browser automation, demo capture, and documentation.

12 skills

dx-automation

Autonomous agent infrastructure: AWS Lambda, ADO pipelines, CloudWatch alarms.

11 skills

dx-core

Estimation

Skill	Argument	Description	Output
`/dx-estimate`	`<work-item-id or URL>`	Analyze ADO/Jira User Story and produce structured estimation — hours/SP, implementation plan, AEM pages, open questions. Posts as ADO/Jira comment.	ADO/Jira comment

dx-core

Requirements (dx-req-*)

Skill	Argument	Description	Output
`/dx-req`	`<work-item-id>`	Full requirements pipeline: fetch work item, validate DoR, distill dev requirements, research codebase, generate team summary. Posts interactive checkbox comment to ADO/Jira (idempotent).	All spec files + ADO comments
`/dx-req-tasks`	`<work-item-id>`	Create child Task work items with hour estimates.	ADO/Jira tasks
`/dx-req-dod`	`<work-item-id>`	Check Definition of Done against ADO/Jira wiki checklist — validates actual deliverables: PR, tests, build, docs, child tasks. Auto-fixes what is possible, creates tasks for the rest.	`dod.md`
`/dx-req-import`	`<path-to-file>`	Validate external (non-ADO) requirements document.	`explain.md`

dx-core

Figma (dx-figma-*)

Skill	Argument	Description	Output
`/dx-figma-all`	`[ADO ID] [Figma URL]` — both optional	Full Figma workflow: extract, prototype, verify. Coordinator skill.	All sub-skill outputs
`/dx-figma-extract`	`[ADO ID] [Figma URL]` — both optional	Extract design context, tokens, screenshots from Figma URL. Screenshot layer analysis, relevance check.	`figma-extract.md`, `prototype/figma-reference.png`
`/dx-figma-prototype`	`<work-item-id>` (optional)	Research project conventions (2 parallel agents) + generate high-fidelity standalone HTML/CSS prototype.	`figma-conventions.md`, `prototype/index.html`, `prototype/styles.css`
`/dx-figma-verify`	`<work-item-id>` (optional)	Visually verify prototype against Figma reference screenshot. Compares, fixes gaps (max 2 rounds).	`figma-gaps.md`, `prototype/prototype-screenshot.png`

dx-core

Planning (dx-plan-*)

Skill	Argument	Description	Output
`/dx-plan`	`<work-item-id>` (optional)	Generate step-by-step implementation plan with status tracking. Optionally invokes `superpowers:brainstorming` for design exploration.	`implement.md`
`/dx-plan-validate`	`<work-item-id>` (optional)	Verify plan covers all requirements, no extras, dependencies correct.	Warnings / OK
`/dx-plan-resolve`	`<work-item-id>` (optional)	Research and fix risks flagged by validation.	Updated `implement.md`

dx-core

Execution (dx-step-*)

Skill	Argument	Description	Output
`/dx-step-all`	`<work-item-id>` (optional)	Execute all plan steps autonomously (step with test, review, commit loop). Loads fix memory, logs fix patterns, promotes proven fixes to `.claude/rules/`.	All steps done
`/dx-step`	`<work-item-id>` (optional)	Execute next pending step from implement.md with integrated test, review, and commit. Optionally invokes `superpowers:test-driven-development`.	Code changes + commit
`/dx-step-fix`	`<work-item-id>` (optional)	Diagnose and fix a blocked step (compile error, test failure, review rejection). Escalates to root cause analysis if targeted fixes fail. Optionally invokes `superpowers:systematic-debugging`.	Fix applied
`/dx-step-build`	none	Build and deploy using config build command, auto-fix errors iteratively.	Build pass/fail
`/dx-step-verify`	`<work-item-id>` (optional)	6-phase verification: compile, lint, test, secret scan, architecture, code review (max 3 fix cycles). Optionally invokes `superpowers:verification-before-completion`.	Verdict

dx-core

Pull Request (dx-pr-*)

Skill	Argument	Description	Output
`/dx-pr`	`<work-item-id>` (optional)	Create pull request via ADO MCP, generate description from share-plan.md. Optionally invokes `superpowers:finishing-a-development-branch`.	PR URL
`/dx-pr-commit`	`[pr]`	Commit changes with ADO work item linking; add `pr` to also create a PR.	Git commit [+ PR]
`/dx-pr-review`	`<PR-id or URL>`	Review a single PR — analyze diff, save findings, post to ADO with optional fix patches + vote.	Findings + ADO comments
`/dx-pr-review-all`	none	Batch-review multiple open PRs assigned to you.	Multiple reviews
`/dx-pr-answer`	`<PR-id or URL>` (optional)	Answer open PR comments with codebase context, detect proposed patches. Applies agree-will-fix code changes, commits, pushes, and replies to threads.	ADO replies + code
`/dx-pr-review-report`	`<PR-id or URL>`	Generate categorized report from PR review — groups by category, tracks patch resolution, posts to wiki.	Report + wiki page
`/dx-pr-reviews-report`	`[—any] [PR URL \| Repo URL \| count]`	Generate categorized reports for multiple PR reviews — batch reports posted to ADO Wiki or Confluence.	Batch reports + wiki

dx-core

Bug Fix (dx-bug-*)

Skill	Argument	Description	Output
`/dx-bug-all`	`<bug-id>`	Full bug workflow: triage, verify, fix. Logs bug patterns, detects component hotspots.	Bug fixed
`/dx-bug-triage`	`<bug-id>`	Fetch bug from ADO/Jira, find affected component, save triage.	`raw-bug.md`, `triage.md`
`/dx-bug-verify`	`<bug-id> [before\|after\|qa]`	Reproduce bug via Chrome DevTools. `before`: confirm bug. `after`: verify fix locally. `qa`: verify fix on QA post-merge.	`verification.md` / `verification-local.md` / `verification-qa.md`
`/dx-bug-fix`	`<bug-id>` (optional)	Generate fix plan from triage, execute, create PR.	`implement.md` + PR

dx-core

Agent Roles (dx-agent-*)

Skill	Argument	Description	Output
`/dx-agent-all`	`<work-item-id>`	Full pipeline: RE, plan, develop, review, PR with checkpoints. Optionally invokes `superpowers:executing-plans`.	End-to-end delivery
`/dx-agent-re`	`<work-item-id>`	RE Agent — analyze story, produce structured requirements spec, post ADO/Jira comment.	`re.json`
`/dx-agent-dev`	`<work-item-id>`	Dev Agent — implement from RE spec, self-check (build/test/lint), commit.	Code changes

dx-core

Documentation

Skill	Argument	Description	Output
`/dx-doc-gen`	`<work-item-id>`	Generate wiki docs from spec files (after development). Reads aem-doc-gen output for Authoring/Website sections. Posts to sprint-specific ADO Wiki/Confluence subfolder.	`docs/wiki-page.md`
`/dx-doc-retro`	`<work-item-id>`	Retroactive wiki docs for completed stories — fetches ADO story, finds linked PRs, searches codebase, generates simplified docs. No spec files needed.	`docs/wiki-page.md`

dx-core

Quality + Utility

Skill	Argument	Description	Output
`/dx-axe`	`<URL> [—fix] [—standard wcag2aa\|wcag21aa]`	Accessibility testing using axe MCP Server — analyze violations, remediate, verify. Requires Docker + Axe DevTools API key.	Violation report, fixes
`/dx-init`	none (interactive)	Configure project — detect environment, generate `.ai/config.yaml`, README, rule templates.	`.ai/` directory
`/dx-adapt`	`[aem-fullstack\|aem-frontend\|frontend]`	Auto-detect project type, structure, and build commands. Saves `.ai/project.yaml`.	`.ai/project.yaml`
`/dx-doctor`	`[dx\|aem\|seed-data\|auto\|all]`	Check health of all dx workflow files across installed plugins — config, rules, scripts, seed data, MCP, settings.	Status report
`/dx-upgrade`	`[dx\|aem\|auto\|all]`	Fix all issues found by dx-doctor — updates stale files, installs missing files, reports manual actions.	Upgrade report
`/dx-ticket-analyze`	`<work-item-id or URL>`	Research ADO/Jira ticket, find all relevant source files.	Research report
`/dx-eject`	`[dx\|aem\|auto\|all]`	Eject plugin assets into local repo — copies all skills, agents, rules, templates so project works without plugins.	`.claude/skills/`, `.ai/ejected/`
`/dx-help`	`<question>`	Answer architecture questions from local `.ai/` docs.	Answer
`/dx-sync`	`[—dry-run] [—parallel] [repo1 repo2 …]`	Sync plugin updates to consumer repos — runs sync script with selected repos and options.	Sync report

dx-core

Decision Support

Skill	Argument	Description	Output
`/dx-council`	`<question>` or context shortcut	Run a decision through an LLM Council — 3 advisors (Critic, Architect, Operator) analyze independently, peer-review anonymously, chairman synthesizes verdict. Context shortcuts: `council the plan`, `council the implementation`. Adapted from Karpathy’s LLM Council.	`council-transcript.md`

dx-aem

Verification + Demo

Skill	Argument	Description	Output
`/aem-snapshot`	`<component-name>`	Baseline component state before development — dialog fields, properties, pages.	`aem-before.md`
`/aem-verify`	`<component-name>`	Check component after deployment, compare against baseline, create test page.	`aem-after.md`
`/aem-fe-verify`	`<component-name>`	Screenshot component in wcmmode=disabled, compare against Figma/requirements, fix loop.	`aem-fe-verify.md`
`/aem-editorial-guide`	`<component-name>`	Open AEM editor, screenshot dialog, write authoring guide.	`demo/` folder

dx-aem

QA + Documentation

Skill	Argument	Description	Output
`/aem-qa`	`<work-item-id>` (optional)	Full QA agent — navigate pages, check rendering/dialogs, screenshot, create Bug tickets.	`qa.json`
`/aem-qa-handoff`	`<component> <id>`	Generate testing steps for QA team, post to ADO. Reuses `/aem-doc-gen` output or creates own test page.	`qa-handoff.md`
`/aem-doc-gen`	`<work-item-id>`	Generate AEM demo docs — find existing pages, create docs page, capture dialog + website screenshots on QA, write authoring guide with Authoring/Website sections.	`demo/authoring-guide.md`, `demo/*.png`

dx-aem

Recon

project knowledge and component discovery

Skill	Argument	Description	Output
`/aem-init`	none (interactive)	Detect AEM structure, append `aem:` section to `.ai/config.yaml`, set up project seed data.	Config update
`/aem-component`	`<component-name>`	Find all source files, AEM pages, and dialog fields for a component (multi-platform, data-driven).	Component report
`/aem-page-search`	`<component-name>`	Find all AEM pages using a specific component.	Page list
`/aem-refresh`	none	Update `.ai/project/` seed data from plugin, external docs repo, or manual sources.	Seed data files
`/aem-doctor`	`[components\|osgi\|dispatcher\|all]`	Check AEM project infrastructure health.	Status report

dx-automation

Setup Sequence

run once in order to provision infrastructure

Automation Setup Order

auto-init Scaffold .ai/automation/

/auto-init

auto-provision AWS resources

/auto-provision

auto-pipelines ADO pipelines

/auto-pipelines

auto-deploy Lambda code

/auto-deploy

auto-lambda-env Env vars

/auto-lambda-env

auto-webhooks Service hooks

/auto-webhooks

auto-alarms CloudWatch

/auto-alarms

Skill	Description
`/auto-init`	Scaffold `.ai/automation/` — generates `infra.json`, `repos.json`, `.env.template`. Prereq check, config questions, data bundle copy.
`/auto-provision`	Create all AWS resources: DynamoDB tables, SQS DLQ, S3 bucket, SNS topic, IAM role, Lambda functions (placeholder), API Gateway.
`/auto-pipelines`	Import ADO pipeline YAMLs into Azure DevOps and set all pipeline variables (LLM credentials, wiki URL, identities).
`/auto-deploy`	Package and deploy Lambda code for DoR, DoD, PR Answer, BugFix, QA, DevAgent, and/or DOCAgent. Wraps `lambda/deploy.sh` with audit logging.
`/auto-lambda-env`	Set Lambda environment variables interactively (ADO PAT, webhook secrets, DynamoDB table names, etc.).
`/auto-webhooks`	Configure ADO service hooks and PR Review build validation policy. WI hooks (project-scoped, hub only) + PR Answer hook (per-repo, all profiles) + PR Review build policy (per-repo).
`/auto-alarms`	Create CloudWatch alarms (DLQ depth, Lambda errors, throttles) and subscribe email to SNS alerts topic.
`/auto-test`	Run a local dry-run of an automation agent against real ADO data (no posts/commits). Verify connectivity and output before going live.

dx-automation

Ongoing Operations

monitoring, health, testing

Skill	Argument	Description
`/auto-deploy`	`[dor\|dod\|pr-answer\|bugfix\|qa\|devagent\|docagent\|all]`	Redeploy Lambda after code changes. Safe to re-run at any time.
`/auto-doctor`	none	Full health check: file integrity, ADO pipeline state, Lambda function state, env var coverage.
`/auto-status`	none	Operational dashboard: DLQ depth, monthly token budget utilization, daily rate limit usage.
`/auto-eval`	`[—all \| —agent X \| —tier2 \| —fixture name]`	Run evaluation framework against test fixtures. Use after changing prompts or agent steps.

Dependencies

Skill Dependencies

How skills chain together.

dx-req chain

dx-req (full pipeline: fetch → DoR → explain → research → share)
GATE: blocks if blocking questions exist
BA checks items → re-run reads state → re-validates

dx-agent-all chain

dx-agent-all
├─ dx-agent-re
├─ dx-plan → dx-plan-validate → dx-plan-resolve
├─ dx-step-all
│    ├─ dx-step (includes test, review, commit)
│    └─ dx-step-fix (targeted fix + escalation)
├─ dx-step-build
├─ dx-step-verify
├─ dx-pr
├─ aem-doc-gen (optional)
└─ dx-doc-gen (optional)

dx-bug-all chain

dx-bug-all
├─ dx-bug-triage
├─ dx-bug-verify (needs triage)
└─ dx-bug-fix (needs triage + verify)