aiocean-plugins

Sharpen Claude Code — patch and recompile Claude's binary for custom use cases (control channels, agent-as-a-service, custom tooling), capture this session's workflow as a reusable skill, consolidate memory, and ship feedback to the marketplace.

9 skills

Understand, plan, debug, and review code as one coordinated workflow — codebase discovery, dependency mapping, implementation planning, root-cause debugging, and pre-merge review.

9 skills

Translate EPUB books into literary Vietnamese end-to-end — upload, pre-translation analysis, chapter translation with consistency checks, quality review, and bilingual export.

9 skills

Curated decision-support reference with semantic search across three domains — 54 general-reasoning mental models (first principles, inversion, second-order thinking, opportunity cost, …), 27 threat-modeling frameworks (STRIDE, LINDDUN, PASTA, OCTAVE, Attack Trees, MITRE ATT&CK/ATLAS, Kill Chain, NIST AI RMF, OWASP LLM Top 10, Kubernetes Threat Matrix), and 137 software-architecture patterns (cloud design, resilience, data, DDD, distributed systems). Each domain routes a problem to the relevant entries, applies each, synthesizes, and stress-tests with a counter-frame. 4 skills: aio-mental-models, aio-threat-models, aio-architect-advisor, aio-architect-reference.

4 skills

Create self-contained HTML artifacts in four genres — report, deck, explorer, editor — routed by what the human does next: read and audit, present, decide, or edit and hand values back. Every artifact is offline (one file, zero remote references), printable (a print stylesheet in every genre), and auditable (claims carry repository, revision, and line anchors, confidence words carry numeric bands, and a parsing validator gates the structure).

4 skills

Drive Jira, Confluence, Google Workspace (Drive/Gmail/Calendar/Sheets/Docs), Tanca HR, and X/Twitter from your terminal — one CLI per service, no context switching.

4 skills

Design and ship UI with confidence — UI/UX guidance, neobrutalism design system bootstrapping, and SaaS analytics dashboard design (chart selection, anti-pattern checks, a11y, storytelling).

3 skills

GitHub, GitLab, and git worktree workflows in one toolkit — manage repos, PRs, MRs, pipelines, and parallel branches without juggling CLIs.

3 skills

Attack-first reference for AI-agent security with local semantic search — how current AI agents (LLM apps, tool-calling agents, MCP servers, RAG/memory agents, multi-agent systems, coding agents) are manipulated and attacked, each vector paired with detection + mitigation. 46 articles in 8 volumes: prompt injection (direct/indirect/cross-modal), tool & MCP poisoning (TPA, rug-pull, shadowing, MCP CVEs), memory & context poisoning (MINJA, AgentPoison, RAG poisoning), multi-agent attacks (ARCJ, worms, orchestrator hijack), jailbreaks (Crescendo, Echo Chamber, many-shot, policy puppetry, GCG), coding-agent supply-chain (Nx s1ngularity, repo poisoning, exfiltration, real incidents), and defense-in-depth (spotlighting, dual-LLM, CaMeL, least-privilege). Every claim carries a source URL and a HIGH/MEDIUM/LOW confidence label. 2 skills: aio-agent-threats (advisor workflow), aio-agent-threats-reference (encyclopedia lookup).

2 skills

Generate shareable Mermaid URLs and auto-build Grafana dashboard diagrams from your codebase — diagrams that travel well, without leaving the terminal.

2 skills

Track AI-agent work in a markdown kanban board committed to the repo — index board with per-task files, an auto-dispatch monitor that spawns sub-agents for matching tasks, and an optional Go terminal UI (aiokan) for vim/mouse board steering.

2 skills

Give Claude Code an event loop. A local HTTP + WebSocket relay plus the Monitor tool let anything outside Claude — a CLI, a mobile app, a webhook, another program, a web page — talk to it mid-task in both directions. Inbound: client POSTs an event, relay prints a MSG:: line to stdout, Monitor surfaces it as a notification. Outbound: Claude POSTs a push, relay broadcasts it over WebSocket. Generic, language-agnostic protocol with a ready-to-run reference relay, plus optional token-gated network exposure over Cloudflare Tunnel, Tailscale, or LAN.

2 skills

Run structured 10-phase research workflows and stand up RAG pipelines on Qdrant — go from a topic to a queryable knowledge base.

2 skills

Design StarRocks tables and tune slow queries — partitioning, bucketing, sort keys, and PK tuning, plus EXPLAIN/Query Profile reading and concrete tuning recipes.

2 skills

Remove AI slop from prose and code. Mechanical detection via a grep-able catalog of English and Vietnamese tells, per-hit triage with false-positive protection, minimal repairs verified by a re-scan loop, and a diff-scoped code pass that keeps behavior unchanged.

1 skill

Push books/docs to an Onyx BOOX e-reader and manage its cloud account (push list, library, notes) API-direct via a self-contained zero-dependency Node CLI — no web UI, no email, no BooxDrop.

1 skill

Extract browser cookies with rookiepy and reuse them for authenticated requests or Netscape cookie exports.

1 skill

Bootstrap a Bun fullstack project in one shot — single-port API + frontend, Vite dev proxy, monorepo layout, and Docker config. Skips files you already have.

1 skill

Turn Claude into a wise teacher that makes sure YOU deeply understand the work an AI agent just did — the problem, the solution, the design decisions, the edge cases, and why it matters. Incremental and mastery-gated: it has you restate your understanding first, fills the gaps at the depth you ask for (eli5 / eli14 / like-an-intern), shows you the real code, quizzes you, and does not conclude until you have demonstrated that you get it.

1 skill

Runs an adaptive, project-aware code-review workflow and carries Google's engineering code-review playbook it applies. The workflow profiles the project, detects its real lint/type-check/test tooling and house conventions, reviews on two tracks (mechanical + semantic 8-dimension rubric), adversarially verifies each finding, and emits an LGTM verdict against The Standard ("improves code health, not perfection"). Also coaches the reviewer and CL-author roles. Source: github.com/google/eng-practices (CC-BY 3.0).

1 skill

Comprehensive BDD field guide — writes and reviews Gherkin with expert quality. Includes Example Mapping, 3 Amigos, anti-patterns catalog, and declarative style guide.

1 skill

Write and lint idiomatic Go end-to-end — a 7-tool lint chain (build, vet, golangci-lint, govulncheck, nilaway, deadcode, race) plus reference-grade patterns for concurrency, errors, generics, gRPC, and production hardening.

1 skill

Remove image backgrounds using RMBG-2.0 alpha matting with despill — works on any image (renders, photos, screenshots). Pairs with the flat-magenta chroma-key technique: render subject on #FF00FF with any image gen engine, then matte for a clean PNG RGBA cutout.

1 skill

Debug iOS apps on physical devices. Build, install, launch, capture logs, pull crash reports, and analyze crashes from the terminal.

1 skill

Author a UserPromptSubmit hook + .jev-hint catalog file that classifies every user prompt against a JSON catalog of {name, useWhen, context} entries via one batched yes/no decision call (OpenRouter Decisions API, model typesafe/jev-1.14), then injects matched entries as additionalContext when confidence clears a threshold. Covers hook mechanics and .jev-hint field-writing best practices from real production tuning.

1 skill

Monitoring and observability strategy, implementation, and troubleshooting. Metrics design (Golden Signals, RED/USE), distributed tracing (OpenTelemetry), alerting, SLOs, dashboards, log aggregation, and tool selection. Includes 7 automation scripts and production-ready templates.

1 skill

Take finished, already-good work (code, prose, plans, analysis, UI) toward its best achievable form. Deletion test first, every edit a challenger the incumbent must lose to, an exemplar-derived bar that rises each cycle, and a fixed-point report that lists what could not be verified.

1 skill

Build production-grade Go TUIs on the charmbracelet v2 stack (bubbletea/v2 + lipgloss/v2) — Elm architecture, restrained styling, and elite patterns for async rendering, layout geometry, keymaps and mode machines, mouse drag and resizable panes, floating modals and command palettes, and test design with visual verdicts.

1 skill

Verify AI-built UI against design via measurement-driven diff — extracts computed CSS and bounding boxes through Chrome DevTools MCP, diffs against Figma reference or a frozen baseline, and feeds numerical deltas back to the agent. No screenshot eyeballing.

1 skill

Build event-driven Go apps with Watermill — Router, middleware, Pub/Sub, CQRS, and integrations for Kafka, AMQP, NATS, Redis Streams, and SQL.

1 skill

Author Workflow-tool scripts that spend effort like it's free and waste none of it. A maximal-effort playbook for multi-agent orchestration: pick an effort tier (quick check → thorough audit → maximize), then scale fan-out width, verification votes, and discovery rounds off it. Composes named amplifiers — multi-modal sweep, adversarial + perspective-diverse verify, judge panels, loop-until-dry discovery, completeness critic, synthesis, classify-and-act routing, tournament/pairwise ranking, and untrusted-content quarantine — plus a task→harness recipe catalog, with the correctness rules (literal meta, determinism ban, pipeline-by-default, budget guard, concurrency caps) that keep the script runnable and resumable.

1 skill

Build XState v5 state machines the right way — design-first planning, params-first typing, and canonical actor patterns for TypeScript and React, with anti-god-machine enforcement.

1 skill

Search YouTube and extract video transcripts using yt-dlp.

1 skill