research
25Command
Start a sprint by researching it. Spawns parallel jira-researcher agents (codebase, patterns, external) and synthesizes findings into RESEARCH.md. Updates STATE.md.
Command
Start a sprint by researching it. Spawns parallel jira-researcher agents (codebase, patterns, external) and synthesizes findings into RESEARCH.md. Updates STATE.md.
Command
Generate a retrospective for the active sprint, a specific sprint, or a date range. Opt-in — never auto-fires.
Command
Review the diff for the active sprint or a given branch/range. Spawns jira-reviewer; reports findings by severity. Does not fix.
Command
Build a static HTML review site from all sprints in .jira/sprints/ and serve it on a local port. Default serves every markdown file in every sprint; filter with --only. Extra markdown dirs (RFCs, specs) can be appended as sections with --include.
Command
Triage the active sprint's plans into independently-grabbable GitHub issues using vertical slices. Adapted from mattpocock's to-issues pattern. Uses caveman lite.
Plugin Claude Code
Per-project workflow workspace. Hosts .project/.jira/.warden state in a private /.studio/ git repo and symlinks each project root into its workspace slug. No tool/skill modifications — consuming plugins see expected paths.
Command
Adopt memory fragments from the studio workspace archive into local Claude memory. Opt-in, per-fragment confirmation via AskUserQuestion. Backs up local on overwrite. Never auto-triggers.
Command
Archive a completed project's workspace — git mv /.studio/ into /.studio/archive/, then clean up the project root (symlinks, .workspacerc, session hook) if a checkout is present. Never deletes workspace content.
Command
Migrate a legacy project into a studio workspace — untracks committed state, then performs standard setup.
Command
Re-sync a studio-initialized project. Re-reads studio.yaml, re-applies the managed .gitignore block, validates all symlinks resolve into the workspace slug, reports drift. Idempotent and non-destructive — never moves project state or modifies the workspace contents. (BASE behavior for P1; memory archival lands in P3.).
Plugin Claude Code
Frontend design skill with 17 steering commands for impeccable UI/UX. Based on Impeccable (https://impeccable.style).
Command
Make interfaces production-ready: error handling, i18n, edge cases, performance optimization, responsive adaptation. Replaces harden + optimize + adapt.
Command
Final quality pass: alignment, spacing, consistency, design system adherence, typography, interaction states, and code cleanup. Replaces polish + normalize + distill.
Command
Improve UX details: copy clarity, onboarding flows, motion/animation, and moments of delight. Argument: copy (improve text), onboard (first-time UX), motion (animations), or delight (personality and joy). Replaces clarify + onboard + animate + delight.
Command
Restructure interfaces: responsive adaptation, simplification, or component extraction. Argument: responsive (cross-device), simplify (reduce complexity), or extract (design system extraction). Replaces adapt + distill + extract.
Command
One-time setup that gathers design context for your project and saves it to your AI config file. Run once to establish persistent design guidelines.
Command
Adjust visual intensity: make designs bolder, quieter, or more colorful. Argument: bolder (amplify safe designs), quieter (tone down aggressive designs), or colorize (add strategic color to monochromatic designs). Replaces bolder + quieter + colorize.
Skill Claude CodeCodex
Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, artifacts, posters, or applications. Generates creative, polished code that avoids generic AI aesthetics.
Plugin Claude Code
Acceptance test playbook. Three runtimes (bash, hurl, agent-browser), three modes (design, run, triage). Project-coupled, sprint-agnostic. Composes with jira sprint features/ and /jira:flow.
Command
Design a warden acceptance plan through guided spec gathering. Detects existing planning artifacts (GSD, jira sprint, GitHub issue) and reads them as the spec source, or walks the user through adaptive Q&A. Self-bootstraps .warden/ on first invocation.
Command
Health check for the project's .warden/ setup. Surfaces engine drift, lib gaps, missing runtimes, malformed audit trail, gitignore mistakes, plan-format issues. Reports findings; asks the agent to walk the user through fixes interactively.
Command
Produce a stakeholder-facing summary of a warden run. Reads the per-run summary plus the JSONL audit trail for trend context. WhatsApp-casual tone, leads with shipped vs regressed.
Command
Brief the session from .warden/HANDOFF.md so a fresh agent or returning operator can pick up warden work cold.
Command
Execute warden plans. Thin wrapper over bash .warden/run.sh that surfaces missing-init errors clearly and passes args through.