Recently added

Operate and change an AWS EKS platform: incident triage from symptom to cause, GitOps chart-values changes, and MCP access to Argo CD, CloudWatch, Grafana, Kubernetes and GitHub.

Produce wireframes, prototypes, frontend layouts, technical diagrams, evidence-led charts, HTML reports, interactive explainers, and generated or edited images.

Build agents and their interfaces: write prompts, skills, and MCP servers, delegate across subagents, enforce output schemas, and use separately connected Agent Memory.

Agent Skills plugin for rigorous research nonfiction and research book writing: accessibility support, dictation notes, reading-load triage, prose repair, research intent routing, source discovery, source notes, extraction tables, literature review, argument architecture, chapters, claim traceability, citations, release audits, and proposal support.

Deliberate Codex context compaction with a verified handoff and one-time continuation.

Codex plugin with Editorial, Faithful, and Plain Language Humanizers.

Multi-phase build orchestration with native debugging: assess, plan, execute, review, iterate. Plan-verify gate on Phase 2. Optional UI design gates. Native debug-loop, debugging-memory, and logging-tracer skills provide root-cause investigation without a bundled MCP server. Orchestrator owns when-to-fire; bundled skills own procedural detail; bridges are optional escalation hops to supporting plugins for extended capability.

Evidence-gated review and bounded repair of a concrete Git change scope. Implicitly use only to assess uncommitted changes, a branch or diff, a local ref range, or a PR for material defects, regressions, test gaps protecting changed behavior, or merge readiness. Do not implicitly use for document or generated-output review, output diagnosis, general skill, plugin, or repository analysis, architecture exploration, or planning-only work.

Index an env-configured Windows file tree into a persistent Markdown wiki and answer questions with source-linked evidence.

Read-only macOS operational baseline audit workflow for local-machine diagnostics.

Svelte Development gives Codex a remote-MCP-backed operating contract for serious Svelte 5 and SvelteKit architecture, implementation, and verification work.

Prompt Architect turns rough intent into copy-ready prompts with risk, evidence, authority, output, verification, and execution-topology controls.

Audit, verify, reconcile, and update Codex memory artifacts across project and global scopes through evidence-based, approval-gated procedures.

Live Research gives Codex a disciplined workflow for verifying change-sensitive claims against current authoritative evidence.

Keep changelogs and architecture decision records aligned with completed work, release preparation, and explicit decisions through evidence-based, infrastructure-aware procedures.

Audit and propose repository-first Prettier configuration with explicit approval before any repository mutation.

人間が pin した目的を、build 手順なしで、あらゆる Claude Code セッションへ運ぶ。SessionStart hook が frame・未処理の baton・この unit の repo ごとの corpus fence（drift-intra / drift-inter / working-delta / unpushed / checkpoint-stale）・open-todo 数・escalation 数を注入する。UserPromptSubmit hook が、最初の turn が動く前に一度だけ baton を読む手順を述べる。PostToolBatch hook が、セッションの足元で corpus が動いたときに事実を述べ直す。PreCompact hook が、沈黙の自動圧縮を一度だけ差し止め、代わりに baton を著せるようにする。skill は handoff（/bearing:handoff r|w）。aim の規律は project ごとの opt-in である —— /bearing:setup-aim が実行した project の `CLAUDE.md` の末尾へ marker 付きの法を差し込み、`.claude/skills/aim/` へ aim skill を置く。置いた後はどちらもその repo のものであり、plugin は追随させない。hook も面も marker だけを読む ∴ 採用を宣言していない repo では、corpus を持っていても 1 byte も出さない —— 4 つの面すべてが同じ述語を通る。corpus が在ることは使っている証拠であって、この機構を通したいという宣言ではないからである。例外は未読の baton 1 つで、handoff は aim に何も依存しない。採用していない repo では statusline の 2 行目だけが、corpus を見つけたことを 1 行述べる —— 黙る機構は自分の存在を告げられないからである。marker は HTML コメント ∴ 消費者の context には乗らず、版と本文 sha を運ぶので「版が古い」と「人間が編集した」を別物として 述べられる。corpus の在り処は既定 `docs/aims/` で、--dir で project が宣言できる。`bin/statusline.mjs` は statusline を描く実行ファイルとして供給される —— plugin は statusLine key を宣言できない ∴ 装着は人間の act として残るが、`/bearing:setup-statusline` が人間の home（`~/.claude/`）へ薄い shim を置いて settings.json の 1 行を書く。shim は走るたびに install record を読んで今の版へ橋渡しする ∴ 1 行に version が入らず bump で腐らず、載っていないときはそう描く。 aim の面（browser で開く 1 枚）は /bearing:setup-surface が人間の home の固定 path へ置く —— browser は開かれた HTML の版を解決できない ∴ 橋渡しではなく複製であり、--check が今のかを答える。build 手順も server も daemon も無い。 cache の複製は、bearing の checkout の中に居ると分かったとき working tree 側へ委譲する —— 開発中は working tree、他 project では cache が自動で選ばれる。

Kaiju Ego Avatar: Token-saving AI persona with Gluttony and Aura skills

A governance layer for agentic software development: compose a small set of invariants onto your project and keep your agents honest.

Axon research, ingestion, retrieval, and operator skills for an existing service.

Frontend coding conventions plus the workflow around them. One fe-coding skill detects the stack (Next.js App/Page Router, React Native Expo/CLI, Next.js API + Prisma) and applies the _modules architecture, component hierarchy, Link-only navigation, function minimalism, visible empty states and Zod/React-Hook-Form validation. Workflow skills cover project setup, Figma-to-code, ticket workflow, BA/PO ticket writing (requirement → templated ClickUp ticket), mobile release notes, deployment checks, spec-driven (OpenSpec), and capturing corrections as rules. Each project keeps its own live, committed copy of the rules (edit it, review the diff, PR it back) and can register the other repos of its system so cross-repo contracts are read from the real files instead of guessed. Bundles the context7 and Framelink Figma MCP servers so they load on install.

mARC (Multi-Agent Relay Control) — an IRC/vaporwave-styled Claude Code plugin that packages a full agent team (tech-lead operator + engineer, sre, design, security, review, research specialists) as a reusable, cross-repo team. @techlead convenes the specialists in the channel.

Production-discipline rules for a software project, each carrying the real incident that produced it — the verifiable-gates skill, rendered from rules.yaml; nine of the rules are decided mechanically by the bundle's own scanners once a project installs it (pip install verifiable-gates; python -m verifiable_gates.install). Installing this plugin lands the whole repository and adds a PostToolUse hook that runs the installed doctor after every Edit or Write — off by default; on with VERIFIABLE_GATES_AT_EDIT=1 in the project's .claude/settings.json under env. It reports and refuses nothing.

Personal cross-tool agent plugin: skills and MCP servers for Claude Code, Codex, Kimi Code and pi.

Runs your coding agent as a pipeline of specialist agents rather than one: architecture, product, implementation, QA, security and deploy, with an independent model reviewing each stage before the next begins.

super-fr base: superpowers-wrapped planning, isolation, and goal-to-PR pipeline with GitHub tracking

super-fr dispatch: queue phases to autonomous runners (VibeKanban adapter included) with per-PR quality gates

LINE platform development skills — Messaging API, LINE Login, LIFF, MINI App, and Notification Messages. Provides API references, code patterns, and expert guidance for building LINE integrations in Japan, Thailand, and Taiwan.

Design engineering for AI agents — animation timing, transitions, typography, color, components, layout, accessibility, anti-patterns, sound, SVG, launch films. Plus a SOUL.md identity layer and ten workflow subagents. Distills Emil Kowalski, Benji Taylor, Jakub Antalik (transitions.dev), guidelines.sh, Vercel, Ben DC, DiceBear, lucide-animated, Google Labs design.md, Agentation.

Show 1/8, keep 7/8 below the surface. One-line answers, a four-line ceiling, no summary blocks, and -a for depth.

Execute implementation plans for Stargazer via paired implementer+reviewer agents with direct communication

Read-only capability-based specialist review with validated confidence-scored findings for the Stargazer codebase

Batch-parallel plan execution for the Stargazer codebase — advisor (opus) + implementers (sonnet) per batch, autonomous loop until all tasks done

Adversarial verification skill: prover vs disprover agents argue, then a judge synthesizes a verdict

Controller-led mixed-agent implementation planning with Claude, Codex, and OpenCode

Stepwise design and reconstruction of existing code, with whole-file source versions, incremental sync, and an interactive design reader.

Removes 20+ patterns of AI slop from any piece of writing.

Professional writing skills for AI coding agents: anti-slop editing that makes drafts sound human, and technical content writing with runtime voice calibration.

Codex workflow skills for building brief-driven AI projects, concepts, and teaching demos.

Find Figma components across configured teams and inspect implementation-ready node contracts.

QP (Quality Platform) — 4 skills (qp · qp-plan · qp-run · qp-regress). Plan→Automation→Regression loop for web and desktop apps with zero domain constants. Calls a registered qasp typed capability when one is present (strictest-wins); otherwise drives the target through the session's Playwright MCP (web) or computer-use MCP (desktop). Mobile deferred.

New in 2.0.1: harness-doctor cadence promoted from prose to a SessionStart hook (30-day overdue check no longer silent); portability lint (BSD/GNU, zsh/bash pattern denylist) wired into pre-commit as the shell-layer sibling of the existing lints; branch_claim.sh gains an explicit warning that live-claim count is not a same-tree concurrent-edit count; the public-surface confidentiality scan's fail-open notice (operator pattern override absent) is now a hard-to-miss banner instead of a two-line aside; fh-gate.sh survives a missing package.json instead of dying under set -euo pipefail; identity ① (harness cluster) reclassified 🟢 under a redefined ⓐⓑⓒ bar (reuse is real and callable, FH did not grow to get it, sharpness is preserved via strictest-wins merge) — cross-harness adapters + relay_channel.sh --cap-args node-specific argument channel; §1-a-2's 6-axis verification canon synchronized across the file (a prior draft still said 4 in one place). New in 1.4.74: shipped documents no longer point at files the package omits. 35 such paths were measured on 2026-07-28 — CLAUDE.md instructed consumers to run templates/predelete_check.sh before a destructive op and the file did not ship. 20 more paths added; the 4 that remain absent are enumerated with a written reason each (per-environment registries and operator-private sync). Closing the set once is worth little because it regrows, so scripts/package_coverage_check.sh now fails selfcheck on any new occurrence, known-pair calibrated, and it caught itself on first wiring. New in 1.4.73: the mechanical floor now ships — templates/.git-hooks (pre-commit/pre-push) plus every script they actually invoke, their regression anchors, and the .claude/rules data they read. Until now the package shipped the prose that tells you to run the gates while omitting the gates, and 35 shipped-doc references pointed at files the tarball did not contain. Verified by unpacking the tarball, installing the hooks in a scratch repo and committing through them, not by inspecting files[]. degrade_direction_scan gains shell probes: it collected .sh files that no probe could match and reported them clean (known-positive scored 0/4); git hooks were invisible entirely because the extension test ran against the path, not the basename. The field-propagated templates/ copy had drifted from scripts/ and is now byte-pinned by an anchor. Hub meta-engineering toolkit — 35 skills + 7 agents. New in 1.4.71: SKILL_detail.md brought inside the 4-axis gate — the gate matched the literal `SKILL.md`, which `SKILL_detail.md` does not contain, leaving 27.7% of the skill-spec surface ungated (measured; it leaked twice for real). Fix escalated from a name list to a directory scope after an enumeration sweep found a real uncovered file on its first run; anchored by scripts/gate_pathspec_check.sh (known-pair, wired into pre-commit). One-way mirror sync gains a destination-newer abort + mechanically injected mirror banner after two session cards were silently overwritten, anchored by scripts/sync_guard_check.sh. dialogue-harvest gains Step 4-b cross-corpus provenance (absorbed / held-unused / declined) for single-author corpora, where the original provenance labels were free and measured nothing. frontier-digest arxiv leg category-scoped after a relevance drift the staleness rule structurally could not catch (known-pair calibration recorded). templates/CLAUDE.md hub paths switched to {FH_ROOT} so a copied harness stops carrying one operator's absolute paths into org-visible checkouts. New in 1.4.70: dialogue-harvest (mines AI-dialogue logs — sycophancy strip first, induced-vs-independent provenance labeling, EN+KO known-pair calibration shipped); asset-placement-gate Step 0.6 official-corpora check (Claude Cookbook as Tier-0 consult corpus, judged-flag semantics); context-doctor built-in-/doctor-first sister anchor. New in 1.4.48: phantom-quench + steel-quench gain external frontier anchors (arXiv:2607.02052 package-hallucination; arXiv:2607.02057 prompt-coverage-adequacy); README model-flat claim reframed from a per-release point-curve to structural invariants (operation flattens across tiers; depth tier-order fixed within a generation). New in 1.4.47: onboarding step ① surfaces the Mode D companion-store session-start load in the auto-read salience anchor (previously only in the local binding + rules, so a greeting could skip the load). New in 1.4.46: context-doctor gains a command-output axis — routes to a command-output proxy/hook (rtk) to trim verbose CLI stdout, complementing .claudeignore; risk-gated to token-scarce environments (lossy filtering, off gate-input paths). New in 1.4.41: context-doctor 2026 trigger vocab (context engineering/rot/collapse) + phantom-citation hardening; hub measurement-integrity-checklist (cross-model measurement pre-flight: display-name pin/reps≥3/discriminating probe). New in 1.4.40: install-wizard scaffolds the companion store as a queryable wiki (INDEX + session-start read + Raw/Wiki/Conversation ingest axis). New in 1.4.39: auto-decorrelation (cross-family verifier sidecar recruitment, calibration-gated) + video-ingest (capability-routed video ingestion). New in 1.4.37: corpus-grounding-expander + persona-roster-expander (field-harvested verbatim-relay capability skills). New in 1.3.0: public-surface-audit (git-tracked private-token leak scan), field-harvest Mode B session-end auto-trigger, 4-axis gate scope extension (docs/ + AGENTS.md). New in 1.2.0: pipeline-conductor (4-pipeline gated sweep), return-path-gate (chain closure audit), goal-quench (Stop hook + quality gate), steel-quench Wave 5 (multi-model sidecar challenger), 2-layer architecture docs, YAML validation script. Validated cross-CLI: Claude Code, Codex, Gemini.

Cross-project utility skill bundle — 6 skills + 1 agent. Domain-agnostic and portable: convergence-loop, deliberation, mcp-circuit-breaker, token-budget-gate, ko-tech-writer, preprep + quench-challenger agent.

10-phase autonomous planner — turns business ideas into dependency-ordered planned tickets. Appraisal → Discovery → Architecture → Specify → Review → Consensus → Crosscheck → Epic Gen → Ticket Gen → Completed. Produces Planner Context blocks, Branch Directives for shared epic branches, and initiative epics with planned tickets ready for fleet-controller dispatch.

Fully autonomous Linear ticket pipeline — appraise, implement, verify, and merge with zero user input. Includes 20+ slash commands, state-machine-driven flow control, safety gates, and retrospective analysis. Supports shared epic branches with per-ticket git worktree isolation.

Durable cross-project knowledge tracking — capture ideas, decisions, lessons, and discoveries with low friction, automatic resurfacing, and a dispatchable work stack. Curates over existing systems (claude-mem, Linear, openspec) rather than creating a new silo.

Pre-work readiness gate — assesses business ideas and other inputs against profile-driven dimensions, scores them deterministically, and produces a cryptographically sealed Validated Business Intent document. Any agent can invoke it before acting.

Parent orchestrator above ticket-planner and ticket-auto — dispatches planned tickets, monitors pipeline health via 14 detection engines, manages epic branch lifecycle (create, sync, GC), aggregates execution feedback. Bash-only detection + intervention; no Claude agents.

Reconcile the MotifPath Project Index page in Notion at session close.

Strategic Product Manager thinking partner for the MotifPath music education platform.