Toasts

Verbatim Jev-guided compaction for Claude Code sessions.

Trim long Bash output with Jev before the model sees it (Claude Code plugin)

Lossless context compaction for Claude Code: old tool output moves to local files and comes back unchanged by id, or by what it is about with a Jev key.

Lets Claude compact its own context at a step boundary it picks, with its own instructions, then resume the work.

Tool-call guard rules (parser-based). On by default: blocks malformed bash and find rooted at /. Opt-in: Monitor, find without -xdev, pgrep/pkill pitfalls, until-grep loops, sudo, run0 confirmation

바탕화면에 사는 픽셀 펫 클로: Claude가 생각하고, 도구를 쓰고, 허락을 기다리고, 일을 마칠 때마다 반응합니다. 마우스로 끌어 옮길 수 있습니다

ターン終了時のコンテキスト使用率やプロンプトキャッシュの失効を見て自動で Compaction し、要約をテキストに保存する

/goal の内容をタスク分解・TODO化・実行順序付けし、各タスクを SubAgent に委譲させる

変更範囲の大きい Issue の実装を、gh stack の Stack PR に分割して自律的に進めさせる

SubAgent の起動時に、タスクに合う model と effort を ollama の決定モデル (tev1:4b、不調なら nimble) に判断させて振り分ける

Warns when another Claude session on the same repo already edited the file being changed, or claimed the same package version.

logs merges, releases and deploys across sessions and drafts build-in-public posts

Sends a desktop notification when a long turn finishes, so you can look away while Claude works.

Blocks Write, Edit and Bash calls that would put an API key, token or private key into a file or command.

Masks API keys and tokens in tool results before Claude reads them, and blocks a git push that would leak secrets or home paths to a public repo.

/snap: drag a box on screen with Spectacle and the screenshot goes into your next message.

Verifies signatures on the instruction files Claude Code loads, meaning CLAUDE.md, skills, and agent definitions, before they reach model context. It re-checks a skill bundle before the Skill tool runs it. Sigstore keyless, offline verification against a pinned trust root, no telemetry.

Skills and guard hooks: ai-code-cleanup, interrogate, shape-task, verify-change, setup, a shared-worktree git guard, and a guard against reading API-key config files

Shrinks the context without losing the conversation: bulky old tool results become one-line stubs, every message stays word for word.

Compacts the conversation while you are away, a few idle minutes in and before the prompt cache goes cold, instead of in the middle of your work.

Notices when Claude is stuck repeating itself (same failing call, same output, edit flip-flops, failure streaks) and tells it to change approach.

A chime, a toast and an optional webhook when Claude finishes a long turn or needs your permission.

Ask before guessing: when a prompt looks ambiguous, Claude asks up to three clarifying questions first. Off until /clarify on.

Inline #tokens that attach live context to a prompt: #diff, #staged, #log:20, #tree, #todo, #errors, #file:path:10-40, #url:https://...