Prompt

Replies that point at code: ask with /code-reference (a question, or a PR to review) and each place the reply names opens in a code pane beside it, the prose linked to it and boxed under each section. Only the command triggers it. Drawn the same in the terminal and the desktop app. PRs are read with gh, a GitHub MCP server or git, and checked out into a review worktree.

Every subagent Claude starts waits for your approval: approve, refuse, approve all for the turn, change its model (suggested by the task's complexity) or add a note to its task. A pane lists the agents waiting, the full task and recent decisions, the same in the terminal and the desktop app.

Hands a long session over, to a fresh one or by compacting, before the context gets too big

A pane beside the transcript that fetches one Jira issue by key and lets you attach it to your next prompt.

A pane beside the transcript that shows the model's open interview questions and submits every pick as one prompt.

Development standards: look before building, test what changed, pass the repo's checks, keep the rules in view.

Keeps AI coding agents inside your project's rules, from idea to merged code: spec design, lanes, layer rules, protected paths with an approval log, spec-check and design gates, proof-before-finish checks, dispatch, review, build status and CI verification, configured per project from its docs

Harness opinionado de Claude Code para construir aplicaciones serverless en Azure con Event Driven Architecture, Event Sourcing (Marten + Wolverine), Postgres y .NET 10. Incluye pipelines TDD, IaC, tooling, scaffolding y agentes/skills especializados.

Session dashboard: a /tether pane for a working-on note, context, the assumptions Claude is making, the work it puts off, decisions, needs-you tasks, subagents and cost. Every section folds.

A brief view of the repo's BACKLOG.md: what's next, and how many items are to-do, new, parked or blocked

dima's command center above the prompt, one row: orbit 🪐 (the session's asks, ticked in the board instead of pasted, and its 🌔 phases), the 5h and ctx meters, afk (one switch every session obeys), keep-hot 🔥, the ⏰ waker, and the /board fleet board

A pane on the right listing the decisions Claude needs from you, each with its recommendation and Approve / Chat buttons

Claude says the tests pass. Receipts checks. Flags success claims that no check backs up.

After each response, a panel above the prompt with what needs you, errors, the next action (with its task id), open to-dos and heads-ups.

/handover writes a handover note from the conversation (served from the prompt cache), saves it under ~/.claude/projects/ /handovers/ .md, clears the session and continues from the note in a fresh context. Suggests it when context passes a threshold.

In a worktree session, tells Claude up front which commands the worktree guard refuses (heredoc and inline-script edits, multi-line scripts, cd to computed paths), so it uses Edit and plain commands instead of retrying.

Click-to-copy chips above the prompt for the commands, ids, paths and URLs in Claude's last replies

Ather Automata: what needs you and what is next in every S2 session and web app repository (Plan, Build, Prove, Ship), a newcomer tour, autonomy windows, and guards that catch false build passes and lost merge edits.

A configurable status bar for Claude Code — band, status line, pane, and toasts driven by one config; configured with npx @sushant-kum/ccstatus or /ccstatus.

Git only when you asked: commits, pushes, pull requests, merges and destructive git commands run only when your latest message (or /ship, /kickoff (also as /progress-pane:kickoff), or your answer to Claude's question) allowed them; everything else is refused with a note telling Claude to ask. No UI.

Mission control for Claude sessions, in a Notebook look: a band and /progress pane in every session (model, effort, tokens, context, agents, gate, git), and with a worklog a validated worklog tool Claude updates at every step (done only after a green gate), facts the mod sees for itself (gate runs, commits, branch, agents, context and cost), drift hints where the two disagree, a band above the...

One mod for Claude Code, Windows and CJK first: pasted-image and text previews in any terminal, themed replies with CJK-correct tables and charts, long edit diffs folded, unfinished tasks carried into the next session, and guards for multi-agent runs and Traditional Chinese writing

Watch Tower: a pane to follow a Claude Code session at a glance (subagents and what they are doing, context, cost, quotas, prompt cache countdown, attached plan's position, off-plan guard), with an optional ASCII companion

Paints Markdown onto Claude Code's prompt box as you type: fenced code becomes a syntax-highlighted card (even before you close the fence), and inline code, bold, italic, lists, task lists, quotes, tables, links and headings are styled. Paint only, your text is never changed.