Dev Rewind

    1

    Plan your day and reflect on your work through a Git-based conversation.

    Free

    1 useSecurity scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Dev Rewind

    Example session with this skill installed

    Representative request: Help me plan today from the last 24 hours of Git activity in my selected repository.

    Test context: synthetic repository, morning flow, with follow-up answers confirming unfinished message changes, planned tests, and no blockers. This is a summary of the multi-turn test inputs, not a verbatim single prompt.

    • Read your context and instructions
    • Compiled the dev rewind

    Recorded test output

    The complete final response below was captured on 2026-10-15 in a synthetic repository using Codex CLI 0.162.1 on WSL Ubuntu with gpt-6.1-sol. It is a local multi-turn test, not an Agensi execution or real project activity.

    朝会の最終レポートです。

    対象: demo repo / master
    期間: 2026-10-14 06:24:32 ~ 2026-10-15 06:24:32(UTC)
    取得範囲: 現在のHEADから辿れる全authorの非マージコミットと作業ツリーの変更状況。GitHub照会なし。

    確認できた作業

    • 8c54474 — 合成デモのベースライン追加。

    作業中

    • 架空デモのメッセージ調整。未完了(ユーザー確認済み)。
    • demo.txt: staged・unstagedそれぞれ1行追加、1行削除。
    • 未追跡: notes.txt。本文は未読。

    今日やること(予定・未完了)

    1. デモのテストを追加する。
    2. その後、メッセージ調整を完成させる。

    ブロッカー

    • なし(ユーザー確認済み)。

    保存・通知は行っていません。

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Dev Rewind turns your selected Git repositories into a conversation about today's plan and what you learned from your work.

    Use it for morning planning or an evening retrospective. It asks one question at a time, separates recorded Git activity from your intentions, and helps you turn the discussion into a concise Markdown report. The default conversation is in Japanese; you can request English.

    The core workflow uses Git. Authenticated GitHub CLI access can optionally add pull request and issue context, but is not required for the Git-only workflow. Dev Rewind is an instruction-based skill, not a standalone app or MCP server. It does not require the separate claude-hurikaeri-mcp npm package.

    Saving reports, exporting HTML, searching saved history and sending notifications are optional operations that require an explicit request and the necessary tools in your agent environment. These are not automatic background actions.

    Requirements and privacy

    Your agent needs access to Git and the repositories you select. Repository information such as commit subjects and filenames is processed by your AI host; this is not an offline workflow. Your provider's data policies and any usage charges still apply. Review which repositories you make available.

    Validation

    The morning and evening conversational workflows have been tested with a synthetic Git repository on WSL Ubuntu using Codex CLI 0.162.1 and gpt-6.1-sol. This does not establish support for every host. Native Windows Codex, Claude execution, optional save/export/notification/history operations, and Agensi delivery have not been verified in that end-to-end test.

    Package

    Includes SKILL.md, a README, a report template, optional-operation references, a synthetic example and the MIT License. No executable scripts, bundled third-party libraries or npm dependencies are included.

    License: MIT. Copyright (c) 2026 t1k2a. Source: https://github.com/t1k2a/claude-hurikaeri

    Validated only for core morning/evening conversations in a synthetic repository on WSL Ubuntu with Codex CLI 0.162.1 and gpt-6.1-sol. Agensi delivery, native Windows Codex, Claude execution, optional save/export/notification/history workflows and multi-repository execution remain unverified in this E2E test. Git facts do not establish task completion; user confirmation is required.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Requires

    Git

    Permissions

    Terminal / Shell
    Read Files

    Core: shell access for read-only Git commands in user-selected repositories, and read access to skill templates/reference files. Optional GitHub context requires authenticated gh and network access. Optional saving/HTML export writes files, and optional notifications use network access, only on explicit request with host tools. No GitHub account is required for the Git-only workflow. These declarations describe tool usage; they do not grant host permissions.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    It uses Git records as a starting point for a conversation about plans and reflection. It asks for confirmation and distinguishes recorded changes from unfinished work and intended tasks. No productivity improvement has been measured.

    Core morning and evening flows were tested on WSL Ubuntu with Codex CLI 0.162.1 and gpt-6.1-sol using a synthetic repository. Agensi delivery and other hosts have not been verified.

    The core workflow is designed to collect Git information without editing or committing code. Working files, staged content and HEAD remained unchanged in the recorded test. Optional report saving or export requires an explicit request. Host permissions still control execution.

    No for the Git-only workflow. Optional pull request and issue context uses the host’s authenticated GitHub CLI. Do not paste credentials into the conversation. Repository information is processed by your AI host under its data policies.

    The skill accepts multiple repository paths. The recorded end-to-end test used one synthetic repository; multi-repository execution has not been verified in that test.

    Six files: SKILL.md, README.md, default-template.md, references/optional-operations.md, examples/synthetic-session.md and LICENSE. The source is MIT-licensed. No bundled executable or npm dependency is included.

    New in Productivity