ahmadhammoud56/MOAA

記録された最新のソース活動: 2026年11月2日 06:41

SkillsMP カタログ更新: 2026年11月9日 14:22

収集済み skills: 36

GitHub スター: 5
GitHub フォーク: 0

AI アシスタントでインストール

このプロンプトをコピーして、使用中の AI アシスタントに送ってください。

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/ahmadhammoud56/MOAA.

リポジトリの README

README の元ファイル：README.md

README の元ファイルを読む

README 全文を表示

# MOAA: Memory, Orchestration, Agents, Automation [![CI](https://github.com/ahmadhammoud56/MOAA/actions/workflows/ci.yml/badge.svg)](https://github.com/ahmadhammoud56/MOAA/actions/workflows/ci.yml) [![Release](https://img.shields.io/github/v/release/ahmadhammoud56/MOAA)](https://github.com/ahmadhammoud56/MOAA/releases) [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE) **Long-term memory, a context that never runs out, honest cost meters and a team of agents for Claude Code.** MOAA (Memory, Orchestration, Agents, Automation) is everything around the model that makes long work hold up. **Strata** keeps a layered memory of every session, searches it by keyword and by meaning, hands a full chat over to a fresh one before the context runs out, and keeps an Obsidian second brain. **Mods** show the prompt cache, goals and context live. A **Specialist Council** of 60 role agents works a problem in parallel and cross-checks the results; in the **Agent Arena** agents compete under a blind judge, and the ones that keep winning are **forged** into permanent council members. The name used to stand for "Mother of All Agents"; it grew past agents. ## Try it in Claude Code ``` /plugin marketplace add ahmadhammoud56/MOAA /plugin install moaa-core@moaa ``` Restart Claude Code, then: ``` /specialist-council:specialist-council Is 221 prime? Factor it. /agent-arena-mode:judge --quick -- Write fib(n) with fib(1)=fib(2)=1 and give fib(20). ``` `moaa-core` installs the council, the arena and moaa-catalog. For the whole setup (about 90 pinned third-party plugins, standalone skills and MCP servers), use the installer: ```bash git clone https://github.com/ahmadhammoud56/MOAA.git && cd MOAA ./install.sh # macOS / Linux / Git Bash (Windows PowerShell: .\install.ps1) ``` ### Staying up to date MOAA updates itself, whichever way you installed it. Once a day, at session start, moaa-catalog checks GitHub for a newer release and installs it in the background, with no window: - **Clone install** (`install.py`): the clone moves to the newest release tag (never unreleased work) and reinstalls with your profile and mode. A clone with local changes or commits of its own is left alone and you are told why. - **Marketplace install** (`/plugin marketplace add`): Claude Code refreshes the marketplace and each installed MOAA plugin. The next session says `MOAA updated itself to X` once; restart Claude Code to load it. The log is `~/.claude/moaa-catalog/self-update.log`. Off: `python plugins/moaa-catalog/catalog.py auto-update off` or `MOAA_AUTO_UPDATE=off` (you still get the one-line notice); `MOAA_NO_UPDATE_CHECK=1` stops the check too. By hand: `python scripts/install.py --update` (clone) or Claude Code's own `/plugin` > Marketplaces > moaa > Update. To get an e-mail per release, use Watch > Custom > Releases on the GitHub repository. New here? The [5-minute guide](docs/GETTING_STARTED.md) walks through every part, and `/moaa-catalog:moaa-help` answers "what can MOAA do?" from inside Claude Code. ## What a council run looks like ![Replay of a real Specialist Council run: council.py resolves each role's model and effort, two agents run in parallel, the coordinator cross-checks them and returns 221 = 13 × 17](docs/demo/council-run.gif) A real run, replayed from its transcript and compressed from 44 s. The analyst factors 221 with Fermat's method; a forged arithmetic checker (an arena agent that scored 97 and 95 in two validated runs) uses trial division; the coordinator checks that they agree: ``` 221 is not prime. 221 = 13 × 17. Both workers agree. Check: 13 × 17 = 221, and 15² − 2² = 221. 13 and 17 are prime. ``` That loop is what MOAA is built around: ``` task ──► Specialist Council (60 roles, per-role model + effort) ──► verified answer ▲ │ │ forge (≥2 validated runs, every ▼ │ score ≥ 85, spread ≤ 10) Agent Arena (8 modes, judge + validator) └──────────── library of winning agents ◄─┘ ``` An arena run, replayed from its transcript: three contestants work in parallel, a blind judge scores them and a fresh validator checks the winner (real run, 233 s, compressed): ![Replay of a real Agent Arena judge run: three contestants write is_leap(year), an Opus judge scores them blind, a validator checks the winner](docs/demo/arena-run.gif) ## What's inside | Plugin | What it does | |---|---| | **Specialist Council** | 60 role agents (debugger, statistician, security reviewer, FEM specialist, …) with per-role model and effort presets ([MODELS.md](MODELS.md)). Split, ensemble or hybrid teams, one integrated and verified result. | | **Agent Arena** | Eight modes: judge, tournament, league, swarm, cascade, red team, evolution and synthesis. Every judged result goes to a local library; proven agents are forged into the council. | | **War Room** | Several real Claude Code sessions on one big job, with a task board, review gates on a different model, a merge queue and a live local app. [Zero-token demo](docs/GUIDE.md#try-it): `python plugins/war-room/warroom.py demo`. | | **Allies** | Hands bounded tasks to Codex, Gemini, OpenCode (free zero-retention models, no sign-in), Kimi, DeepSeek or a local OmniRoute gateway, each on its own account. Read-only by default; writes land on their own branch. | | **Team** | People on separate Claude accounts work on one git repository together: a board on its own branch with task claims (never twice, even in a push race), path locks, dependencies, handoffs, messages and a decision log; each session is briefed on its tasks and new messages. No credential or usage sharing. | | **Strata** | Layered long-term memory in local SQLite (raw sessions, notes, project scenes, your profile), recalled at session start. **Project RAG** over each project's files, **virtual context** (evicted turns recalled after handoffs and compactions), **context handoff** near a full context window or when a silent **canary** shows Claude losing its instructions, a **second-brain vault** (quick capture, daily notes, weekly distill into linked pages, Obsidian-friendly), a Karpathy-style **LLM Wiki** (`/strata:wiki`) a **findings graph** with a grounding check (`/strata:graph`), **grounded answers**: every quote checked word for word against the cited note or file, else "Not in the wiki", **meaning search** through a local Ollama model (keyword-only without it) and a read-only **memory viewer** (`viewer.py --open`). | | **prompt-refiner** | Turns every prompt into a clear brief before Claude works on it, asks about ambiguous points with concrete options (or lets a brainstorming skill ask, never both), and hands the brief to every subagent. | | **moaa-core** | The one-line install (council, arena, catalog), a stdlib **cost calculator** (`cost.py`: blended token rates, plan break-even, local and rented hardware, cost per success), plus a guard that steers long inline `python -c` into a script file and `/moaa-core:project-docs`, which audits a project for AGENTS.md, ARCHITECTURE.md, PRD and the rest of the agent doc set. | | **moaa-catalog** | Skills on demand: big skill packs stay installed but disabled, a prompt hook names the one skill a task needs, and a daily check tells you when a MOAA update is out. | | **Mods** | Four Claude Code mods: a **Meters** panel with real bars (context against the strata handoff, prompt cache, goal, canary, memory queue, vault), **Cache Keeper** (when the prompt cache goes cold, what re-warming costs, keep-warm on request), **Goal Meter** (a progress bar for `/goal` from real acceptance checks) and **Collision Guard** (two chats never overwrite each other's files). Status lines on every surface, Remote Control included. | | **More skills** | `/specialist-council:long-task` (brief, progress file, evidence review, resume), `/agent-arena-mode:karpathy-loop` (keep only changes that move a number), `/agent-arena-mode:effort-trial`, `/agent-arena-mode:fold-trial` (one loop vs a council), `/agent-arena-mode:harness-league` (Claude Code vs the allies), and Anthropic's automation recommender. | ### Always on, no commands Nine things run on their own once installed, and each has an off switch: - **Prompt to brief** (prompt-refiner): Claude rewrites each new request as a brief (goal, deliverables, constraints, acceptance checks), asks about anything material with 2-4 options, saves the brief to `.briefs/`, and gives subagents the brief instead of your raw words. Short replies, slash commands and already-detailed prompts pass straight through. Off: `py plugins/prompt-refiner/refiner.py off`, or start a prompt with `raw:`. - **No more compacting** (strata): at about 85% of the context window Claude stops at a safe point, writes `HANDOFF.md` and `CONTEXT.md` to `.claude/handoff/`, and tells you to open a new chat in the project; the new chat picks the work up with your first message (hidden app sessions in the folder leave it alone). If a conversation is compacted anyway, a snapshot is written first. Off: `py plugins/strata/strata.py handoff off`; `handoff threshold 90` moves the trigger. - **Virtual context** (strata): what leaves the context window stays recallable. Every turn, tool results included, is stored before a compaction and when a new chat takes a handoff; later prompts get back the evicted turns that match them. A chat that continues a chain of handoffs, or one long compacted chat (`handoff mode compact`), behaves as if the window had no end. Off: `py plugins/strata/strata.py handoff recall off`. - **Project RAG** (strata): each project's docs, code, notebooks and PDFs are indexed locally (SQLite full-text search, refreshed in the background at session start). When a request clearly matches passages in the project, the best few are added with their `path:lines`; chit-chat and unrelated requests get nothing. Off: `py plugins/strata/rag.py off`. - **Context canary** (strata): Claude ends each reply with an invisible line (`[//]: # (strata:ok)`). Two long replies in a row without it, once the context is 30% full, mean Claude is losing its instructions, so the handoff starts early. Off: `py plugins/strata/strata.py canary off`. - **Meaning search** (strata): search finds notes by meaning, not only by their words, using a local Ollama model (`ollama pull nomic-embed-text`); nothing leaves the machine, and without Ollama it stays keyword search. `py plugins/strata/viewer.py --open` browses the whole memory in a local, read-only page. Off: `py plugins/strata/embed.py off`. - **Vendored plugin updates**: once a day you are told when flightdeck or prismantis has a newer upstream release; `py scripts/sync_vendored.py NAME` copies it in after review, re-applies MOAA's patch and tests it. - **Second brain** (strata): `strata.py capture "idea"` drops a note into a vault (default `~/code/brain`, a local git repo you create with `strata.py vault init`); every chat adds a line to that day's note; `/strata:wiki distill` files the week's notes into linked pages you can open in Obsidian, and project RAG searches them too. Strata never commits or pushes the vault. Off: `py plugins/strata/strata.py vault off`. - **Inline-Python guard** (moaa-core): a long or multi-line `python -c` gets a note to use a script file instead (it never blocks). Off: `MOAA_GUARD=off` or `guard.py off`. - **The right skill** (moaa-catalog): a prompt that clearly needs a skill gets it named, from enabled or disabled packs. - **Update notices** (moaa-catalog): once a day at session start, a newer MOAA release is announced with the command to get it. While a ruflo plugin is enabled, the global `ruflo` CLI is checked every 3 hours (at session start and on prompts, so long sessions too) and installed or updated in a hidden background process, with its native modules built; ruflo's hooks otherwise run `npx ruflo@latest` on every tool call, which on Windows flashes a console window. Off: `MOAA_RUFLO_UPDATE=off`. ![War Room run view: a team of Claude, Codex and Gemini workers, the board, review gates and a needs-attention strip](plugins/war-room/docs/media/run-live.png) ## What it costs - **Council:** every finished worker's tokens go to a local ledger; `council.py cost` and the end of each council answer show them. A run is usually two or three workers. - **Measured, not assumed:** on short tasks with fixed answer keys, Sonnet and Opus alone already scored 100%, and the council matched them at 2.5–6× the cost and 3.5–7× the time ([benchmark](benchmarks/council-vs-solo/README.md)). So the council now answers small checkable tasks with one worker and keeps the team for ambiguous, long or high-stakes work. - **War Room:** far more than subagents. A measured three-worker Haiku/Sonnet run processed 3–4M tokens, mostly cache reads. It has a usage guard and a spend cap; use it for jobs with several independent streams of work. - **Context:** every enabled plugin's skills are listed in every request. `install.py --report` shows what yours cost, and `--mode lean` or `--prune` trims it ([details](docs/GUIDE.md#context-cost-skills-on-demand)). ## Security, in short The installer adds deny rules that keep Claude's file tools away from secrets: `.env` files, `keys.env` (allies' keys), `~/.ssh`, `~/.aws/credentials` and `*.pem`. They do not cover scripts run through Bash, so keep secrets out of the repository too. MOAA installs third-party plugins, and some of them run hooks with your user permissions. Every third-party plugin is pinned to a tested commit, but pinning is not an audit. Your `settings.json` is backed up and merged, never overwritten, and `--uninstall` reverses every change the installer recorded. War Room launches nothing without your confirmation and denies push, publish and delete to its workers. Read [SECURITY.md](SECURITY.md) before `--full`, and use `--dry-run` to preview every command. ## Requirements - Claude Code 2.1.296 or newer, logged in. The council pins `claude-*-5-5` and `claude-fable-5-1` model IDs. - git, Python 3.10+, and Node 20+ (for `npx`). - Windows: if marketplace clones fail with `Filename too long`, run `git config --global core.longpaths true` once. ## Documentation - [Getting started](docs/GETTING_STARTED.md): a 5-minute walkthrough. - [Guide](docs/GUIDE.md): War Room, Allies, Strata, using the arena and forging, profiles, context cost, install options, the marketplace, tests and maintenance. - [Plugin list](docs/PLUGINS.md): all 100 plugins in the `moaa` marketplace. - [MODELS.md](MODELS.md): each agent's model and effort. [CHANGELOG.md](CHANGELOG.md): release notes. - [ARCHITECTURE.md](ARCHITECTURE.md): how the plugins, hooks and data fit together. [AGENTS.md](AGENTS.md): rules for AI agents (Claude Code, Codex, Gemini, OpenCode) working on this repository. ## Author and licence MOAA is created and maintained by **Ahmad Hammoud** ([@ahmadhammoud56](https://github.com/ahmadhammoud56)). The Specialist Council and the Agent Arena are original work; everything else is credited in [NOTICE](NOTICE) and [THIRD_PARTY.md](THIRD_PARTY.md). To cite MOAA, use [CITATION.cff](CITATION.cff). MOAA's own code is released under the [MIT licence](LICENSE). Vendored third-party plugins and skills keep their own licences (MIT and Apache-2.0, listed in [NOTICE](NOTICE)). Plugins installed from other marketplaces are fetched from their authors and are not redistributed here.

リポジトリの README を読み込んでいます…

このリポジトリの skills

分類待ち

未分類

Read and answer questions about the final Paper2Agent manuscript, its supplementary information, figures, and tables.

原文の言語: 英語

2026-10-11

収集済み skill 36 件中 36 件を表示しています。