KeiSeiKit

Deep research on any topic using parallel agents, web search, and cross-referencing. Use when user asks to research, investigate, or deeply analyze a topic, technology, library, or concept. Triggers on keywords like "research", "investigate", "deep dive", "find out everything about". Supports `--angle` presets for common specializations (competitors, design-refs).

Gateway router for web animation work. Picks between /scroll-animation, /motion-design, or /ai-animation via one AskUserQuestion call, then hands off. Does NOT implement animation — the downstream skills remain the source of truth for their respective domains.

Use when running multi-skill pipelines — new-feature, marketing-launch, design-to-code, web-creation, full-audit, rag-setup workflows

Meta-orchestrator — converts a free-text task description into the right artefact(s) (agent, skill, hook, rule, or block) by composing existing KeiSeiKit primitives. Pure-click decision chain except the single intake field. Enriches `_blocks/` over time via Phase 6 block-augmentation — the kit gets smarter with every invocation.

Deprecated alias for /research with the design-refs angle preset. Routes to /research focused on design patterns, visual archetype references, and award-winning site teardowns.

Continuous parallel quality gate during development — 3 backend agents (security/perf/structure) check every significant code change in real-time. Run after writing code, before committing.

Pre-merge/pre-deploy parallel gate — 4 backend agents (security, tests, deps, regression) plus optional 5th frontend-final-gate agent (production build, Lighthouse, axe full, visual diff full) run final checks BEFORE code reaches main. Last line of defense.

Runtime enable/disable of KeiSeiKit hooks via env vars (v0.15.1). Click-only wizard that emits shell `export` / `unset` commands for the user to paste. Supports per-hook disable, profile switch (full / advisory-off / minimal / off), or full re-enable. Does NOT execute anything — user controls their shell.

Scan a project (or scope of projects) and propose candidate agents, hooks, and primitives based on detected stack, CI, deploy, tests, and env surface. Three modes — Full auto, Step-by-step, Full manual — reuse the same scan+propose foundation and differ only in confirm-gate count. Delegates to /new-agent, /escalate-recurrence, and kei-sleep-queue.

End-to-end VPS provisioning — select provider → plan → provision → harden → verify (ssh-check + firewall-diff hard-gate) → handoff. 6 phases, ≥6 AskUserQuestion calls, defensive-only. Stops if either verification primitive fails.

Hub-and-spoke pipeline that produces a production-grade CI/CD plan and scaffolds the workflow files for a new or existing repo — platform choice (GitHub Actions vs Forgejo Actions), build matrix, OIDC-vs-token secrets posture, release automation, and a security gate — via pure-click decisions across five phases. Emits `.github/workflows/*.yml` or `.forgejo/workflows/*.yml`, a secrets-env scaffold (RULE 0.8), and runs `kei-ci-lint` before handing off. Never writes secret values.

Read or write the cross-session mailbox by @id. Send a message to another Claude Code session (`/msg @name text`), read your own inbox (`/msg` with no args), broadcast to everyone (`/msg all text`), list the whole bus, or discover who is reachable. Thin wrapper over the `kei message` jsonl mailbox — messages land in the recipient's NEXT turn via the mailbox-inject hook (pull, not push). Use whenever the user wants sessions/agents to talk to each other.

Defer a hard question, research task, or design comparison to the nightly remote agent (KeiSeiKit v0.12.0 incubation layer). Runs on top of the v0.11 sleep-sync pipeline — user fills one free-text field plus three clicks, task lands in sync-repo/sleep-queue/ and is processed before REM consolidation. Up to 5 tasks per night, 15 minutes each. Pure-click wizard except the single task-description field.

Hub-and-spoke pipeline that produces a production-grade API design plan for a new or evolving service — style (REST / GraphQL / tRPC / gRPC), resource model, machine-readable contract (OpenAPI 3.1 or GraphQL SDL), versioning strategy, rate-limit + auth handoff, and codegen toolchain — via pure-click decisions across six phases. Emits a spec skeleton, SDK pick list, and a per-surface checklist; never writes secrets.

Architectural rules engine — pattern validation, duplication detection, stack compatibility, and design pattern enforcement. Auto-triggers during code review, architecture planning, and when creating new modules. Triggers on "проверь архитектуру", "review architecture", "check patterns", "проверь дублирование", "stack compatibility".

Multi-agent architectural analysis following the Holographic Development Chain. Use when user asks to design architecture, plan implementation, analyze system structure, or says "архитектура", "план", "спроектируй", "как лучше сделать".

Hub-and-spoke pipeline that produces a production-grade auth/IAM plan for a new or existing app — user flows, identity providers, session strategy, authorization model, and threat mitigations — via pure-click decisions across five phases. Emits a scaffolded env-var list, library picks, and a per-threat checklist; never writes secrets.

Deprecated alias for /research with the competitors angle preset. Routes to /research Phase-5 verification wave focused on competitor mapping, architecture analysis, and market positioning.

Use when creating content — research, outline, draft, SEO, format for platform, social promo

Parallel-agent project/feature kickoff — 4 agents design contracts, scaffold tests, review security, validate structure BEFORE any code is written. Prevents 80% of audit findings at the root.

5-phase documentation scaffolder — auto-detect project type, audit existing docs, generate CLAUDE.md / DECISIONS.md / runbook / README / diagrams / CHANGELOG from KeiSeiKit templates. Each phase is a click-driven file in this skill directory.

Import one or more Google Drive folders as private repos into the local Forgejo dev-hub. Wraps the kei-drive-import bash wizard. Click-only path-picker, optional --scan auto-discover. Requires rclone OAuth (one-time) + Wave 45 dev-hub Forgejo running.

Use when fixing a GitHub issue — reproduce, trace root cause, fix, add regression test

Generates AI images using the nano-banana CLI (Gemini 3.1 Flash default, Pro available). Handles multi-resolution (512-4K), aspect ratios, reference images for style transfer, green screen workflow for transparent assets, cost tracking, and exact dimension control. Use when asked to "generate an image", "create a sprite", "make an asset", "generate artwork", or any image generation task for UI mockups, game assets, videos, or marketing materials.

4-phase pipeline for bootstrapping a new multi-agent project — intake, fork skeleton (branch + ledger row + sub-agent spawn), parallel execution with progress aggregation, and per-branch merge ceremony. Implements RULE 0.12 (agent git-model) at project scale. Hub-and-spoke — each phase lives in its own file and is executed in order.

Hub-and-spoke pipeline for installing the logs + metrics + traces triad on an existing service. Decomposes into 5 phases — scale/stack intake, code-side instrumentation, scrape+ship wiring, dashboard import, alert rules. Pure-click except for env-specific values (endpoints, tokens). Reuses `_blocks/obs-structured-logs.md`, `_blocks/obs-metrics.md`, `_blocks/obs-traces.md`, `_primitives/metrics-scrape.sh`, `_primitives/log-ship.sh`.

Use when auditing performance — baseline, profile, identify top 3 bottlenecks, fix, remeasure

Create a personal AI pet persona via interactive wizard. No TOML editing required.

Browser automation via Playwright CLI — UI testing, form filling, screenshots, scraping. Token-efficient alternative to MCP. Use when user asks to test UI, automate browser, take screenshots, fill forms, or scrape web pages.

Use when refactoring code while preserving behavior — checkpoint, extract, test, audit

Hub-and-spoke pipeline that converts "I need a database for app X" into a designed relational schema, a generated first migration, and optional seed/fixture data — via pure-click decisions across five phases. Emits SQL DDL, a kei-migrate-shaped migrations directory, and a library/ORM pick; never writes production secrets.

Session retrospective triage (RULE 0.14). Runs `kei-memory analyze + patterns`, classifies findings, presents them via click-only AskUserQuestion, routes each selected item to `/escalate-recurrence` (rule+hook), `/debug-deep` (bug RCA), or the audit-backlog (log-only). Self-audit is triage, not implementation.

One-time wizard (5) that configures KeiSeiKit sleep layer. Phase 0 picks mode (local-only / remote-only / hybrid); Phase 0b picks local trigger time (6 options incl. Custom HH:MM). Remote/hybrid generate an SSH deploy key, init the memory-repo, write env refs (RULE 0.8), and emit a `/schedule create` command. Local-only / hybrid emit a CronCreate snippet. Pure-click except 2 free-text fields (repo URL in Phase 2, Custom time in Phase 0b).

Prepare a composed Agent-tool invocation via kei-spawn. Click-only wizard — pick role, describe task, set scope. Emits ready-to-paste prompt + subagent_type + isolation + DNA.

Use when generating tests for untested code — happy path, edge cases, error handling

Use when a project needs testing BEYOND unit tests — fuzzing, property-based, load, E2E, or mutation. Five-phase hub-and-spoke pipeline composes the right mix per language × critical path × CI target, scaffolds configs + corpus + fixtures, wires CI jobs, and defines the crash/regression triage workflow. Pure-click: every decision except intake is an AskUserQuestion.

3-wave parallel audit — Wave 1 discovery (4 agents), Wave 2 verification (4 agents), Wave 3 cross-optimization (1 synthesizer). Each wave runs agents in parallel. Combines security review, runtime analysis, quality check, and constructor pattern into one unified workflow with evidence grading and Go/No-Go verdict.

Triggered when Claude detects the same mistake ≥2× in one session, or the user says "опять / again / second time / already said". Interactively codifies the pattern into a rule + hook + wiki entry via option-picker questions. Pure-click flow — no free-text decisions. Writes rule file, generates hook scaffold at user-chosen severity, registers hook in settings.json via update-config. See ~/.claude/rules/recurrence-escalate.md.

Use when building 3D scenes for web — Three.js, React Three Fiber (R3F), Spline embeds, GLTF/GLB loading, scroll-linked 3D, camera animations. Covers scene setup, model optimization, and performance budgets.

Deep debugging using multi-agent analysis and error pattern matching. Use when user reports a bug, error, crash, or unexpected behavior. Triggers on "debug", "fix", "broken", "crash", "error", "bug".