Guards

Keeps subagents small: tells the main agent how to size and split work, asks a growing subagent to checkpoint, and refuses its tools past a hard limit so a fresh agent takes over

Restyles the Claude Code terminal like an ACP chat: prompt and tool cards, inline diffs, a status band, footer stats, a /dash pane and a done toast

Every Claude turn becomes a snapshot: review what changed step by step, undo it, or travel back. Local, zero tokens.

A /git side panel: changed files with stage buttons, AI commit messages, push, and stashes with AI titles

SDD estilo zero-pi dentro de Claude Code, con un modelo del CPAM por fase

Picks the reasoning effort each task needs. Your session's own model assesses each of your first prompts, knowing what each level can do on that model, and moves the level only when it's confident. Then it locks. Subagents get their own level, chosen by the agent that starts them. The footer shows what it's doing, and /er report shows where your effort went.

Zsemle, a pixel-art companion above the prompt (golden retriever, cat, slime, paperclip): watches every usage limit (50/75/90% warnings, pace forecast, resets, API errors, automatic model switch and compaction, cost), stops work at 95%, guards file writes and cheers long turns. English and Hungarian, terminal and desktop.

Rustで実装された基礎的なコマンドの改善版のみを常に使用するように挙動を変更するMod

実装後の差分を、実装者とは別の読み取り専用レビュアー SubAgent に点検させるMod。指摘の形式検証、再レビュー回数の上限、未実施・未解決の明示を行う。任意で、ローカルLLMによる差分の一次点検 (参考データ) も提供する

fd と ripgrep でファイル名・コード内容を検索するツール (find_files / search_code) をモデルに提供するMod

WebFetch を禁じ、代わりに ctxpack で Web ページをノイズ除去した Markdown として取得するツール (fetch_page) をモデルに提供するMod。ローカルLLMによる出典付きの関連段落抽出も選べる

情報収集した内容を回答やファイルに書くとき、出典 (URL) の併記を必須にするMod

ゴールとタスクの状態をローカルに保存し、Compaction や再開のあとに実際のファイルと照合して復元する

Web・Notion など外部由来の内容を参考データとして扱い、取得後の設定変更・外部送信・公開系の操作を要確認にし、送信先 allowlist と秘密検査で外部へ送る内容を制限するMod

リポジトリごとの検証コマンドと受け入れ条件を登録し、実行結果と変更状態の記録に基づいて完了を判定させるMod

書き込みを行う並列タスクに専用 git worktree を割り当て、ベース・担当範囲・成果差分を追跡し、競合と範囲外変更を検出してから安全に統合するMod

PII and secrets are swapped for placeholders (PII_PERSON_1) by regex and a local Gemma 4 before anything reaches Claude; placeholders are restored only where the work stays on this machine (local tool calls) and on screen.

A compact band of your sessions (Claude, Aesop, named desktop sessions and what they spawn): Clawds coloured by model, glowing by effort, each with its current work

Desktop-only pane drawing the session's running subagents as a left-to-right graph

Hands the conversation off once the context passes a threshold (40% of windows over 200k tokens, 50% of smaller ones): a dedicated turn writes the handoff and the durable knowledge it names into an OKF bundle at .auto-handoff/, then the context is cleared and the work continues from the handoff

Session meters above the prompt; /meter opens detailed metrics

Meters above the prompt: the context window, the session's spend and what is left of your limits, a small mascot that walks them while Claude works, and a line under them with an AI tip, a joke or a piece of good news.

Barra lateral estilo Evangelion: cuota, contexto, qué hace el agente, PRs, subagentes y alertas

Playful badges unlocked once ever from real events (tool calls, turns, subagents, tests, git); /achievements lists them