rewelo

io.github.sebs由 GitHub 帳號 sebs 發布v0.7.3更新於 Oct 24, 2026

Prioritise a backlog with the relative weight method: score, rank, simulate and plan tickets

已驗證STDIO僅桌面效率與工作流資料與分析

概覽

AI 產生的概覽

讓助理用相對權重法為專案待辦事項評分、排序、加上標籤並產生報告,資料存放在本機 SQLite 檔案中。

可以做什麼
Rewelo 透過 stdio 提供待辦事項優先順序工具。每張工單依效益、懲罰、估算與風險四個面向評分,優先順序以(效益+懲罰)/(估算+風險)計算,並在整個待辦清單或某個標籤子集內正規化。助理可以管理專案、工單、標籤、關聯與各專案權重,檢視修訂歷程與稽核記錄,產生報告與儀表板,並匯入匯出 CSV 與 JSON。同一個命令列程式也可作為 MCP 伺服器執行。
適用情境
當你希望助理協助排定產品或團隊待辦事項的優先順序、依投資報酬比較工單、用 state 或 feature 等標籤歸類工作,或從本機待辦資料庫產生摘要與健康報告時,適合使用。
執行需求
透過 stdio 在本機執行,可用 npx 從 npm 套件啟動,或使用 Docker 映像。需要一個可寫入的 SQLite 資料庫路徑,由環境變數 RW_DB_PATH 指定(預設是工作目錄下的 relative-weight.db)。使用 Docker 時需掛載磁碟區以保留資料庫。未宣告需要帳號或 API 金鑰。
安裝前請注意
README 說明這是實驗性軟體,不提供支援,也不適合用於正式環境。助理可以建立、更新與刪除專案、工單、標籤與關聯,匯入操作也可能覆蓋資料,因此應把 RW_DB_PATH 指向可以承受變更的資料庫並做好備份。未掛載磁碟區時,容器刪除後資料會遺失。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 rewelo,將其新增到工作區。
  2. 為需要使用其工具的對話啟用該服務。

Desktop only,透過 STDIO。 STDIO 服務會啟動本機處理程序,因此需要 SourceWeft 桌面主機。

其他 MCP 客戶端

參照 儲存庫 中的啟動說明。

README

Rewelo

Warning: This is experimental software. It comes without support and is not intended for production use.

A prioritization tool that uses the relative weight method to rank stories and tickets by different criteria. Built on SQLite (Node's built-in node:sqlite) for embedded, zero-dependency data storage, exposed through a CLI or as a Docker-based MCP server.

How It Works

Each ticket is scored on four dimensions using the agile Fibonacci scale (1, 2, 3, 5, 8, 13, 21):

DimensionMeasures
BenefitValue gained by implementing the story
PenaltyCost of not implementing the story
EstimateResources required for implementation
RiskUncertainty or complexity in implementation

From these, the tool calculates priority at runtime:

Value    = Benefit + PenaltyCost     = Estimate + RiskPriority = Value / Cost

Higher priority means better return on investment. Relative weights normalize scores across the full backlog or within a tagged subset (e.g. a feature). See calculations.md for the full specification.

Tags

Tickets are organized through a flexible tag system using prefix:value pairs. Tags replace fixed fields for state, feature grouping, and any other classification:

  • state:backlog, state:wip, state:done
  • feature:checkout, feature:auth
  • team:platform, priority:critical

Every tag assignment and removal on a ticket is tracked in an audit log, enabling lead time and cycle time calculations from state: tag transitions.

Revision History

The tool maintains full revision history at two levels:

  • Ticket revisions -- snapshots of content (title, description, scores) and tags before each change
  • Tag revisions -- snapshots of tag definitions (prefix, value) before renames or edits

This means you can reconstruct the exact state of any ticket at any point in time.

Data Model

The database schema is defined in db/create.sql and documented as DBML in db/model.dbml.

projects  +-- tickets              (B, P, E, R scores)  |     +-- ticket_revisions     (content + tag snapshots)  +-- tags                 (prefix:value pairs)  |     +-- tag_revisions  +-- ticket_tags          (current assignments)  +-- ticket_tag_changes   (audit log: added/removed)  +-- ticket_relations     (blocks, depends-on, relates-to)  +-- ticket_deletions     (deleted tickets, for project diff)  +-- weight_configs       (per-project B/P/E/R weights)

Tech Stack

  • SQLite -- embedded database via Node's built-in node:sqlite, no server and no native modules required
  • TypeScript -- CLI application and business logic
  • Commander.js -- command-line interface
  • Docker -- containerized deployment, also usable as an MCP server

CLI Commands

rw project create|list|delete|history|diff                                    Manage projects, view history and changesrw ticket  create|list|update|upsert|delete|history                                    Manage tickets and view revision historyrw tag     create|assign|remove|delete|list|rename|log                                    Manage tags and view audit logrw relation create|remove|list|list-all                                    Ticket relations (blocks, depends-on, relates-to)rw config  weights                  View/set/reset per-project B/P/E/R weightsrw calc    weights|priority         Relative weights and weighted priority calculationsrw export  csv|json                 Export project datarw import  csv|json                 Import project datarw report  summary|group|distribution|health|times|dashboard|event-log                                    Reporting, HTML dashboard and event streamrw serve                            Start MCP server (stdio transport)

Global options: --db <path>, --json, --csv, --quiet (and --no-color, accepted for compatibility: rw never prints colour)

Multi-Project Support

The tool is designed around projects as the top-level boundary. Tags, tickets, and all calculations are scoped to a project, so multiple teams or products can use the same instance independently.

Releasing a New Version

Four GitHub Actions workflows in .github/workflows/ cover building, releasing and publishing:

WorkflowRuns onDoes
ci.ymlpushes and pull requests to mainType check, build, test, and a smoke test of the Docker image
release.ymlpushing a v* tagChecks the tag matches package.json, tests, and creates the GitHub release with the npm tarball, SBOM, OCI image and changelog; pushes the image to ghcr.io/sebs/rewelo
publish.ymlmanual (with a version)Publishes that tagged version to npm via trusted publishing, with provenance, then lists it in the MCP Registry as io.github.sebs/rewelo
pages.ymlpushes to mainBuilds the website from site/ and the docs, and deploys it to GitHub Pages

To release:

  1. Bump the version, which commits it (with server.json, the MCP Registry entry, set to the same version) and creates the matching tag:
bash
npm version patch   # 0.1.0 → 0.1.1 (or minor, major)
  1. Push the commit and the tag; release.yml builds the release:
bash
git push --follow-tags
  1. Publish to npm: run the Publish workflow with the new version (Actions → Publish → Run workflow), or gh workflow run publish.yml -f version=0.1.1. Once npm shows the version, the workflow lists it in the MCP Registry too (mcp-publisher, logged in with the workflow's GitHub OIDC token, so without a secret). Prereleases such as 0.2.0-beta.1 are published under the next dist-tag, and a backport older than the current latest (0.5.2 after 0.6.1) under backport, so npm install rewelo keeps getting the newest stable version.

One-time setup: add sebs/rewelo with workflow publish.yml as a trusted publisher of the rewelo package on npmjs.com, and set Settings → Pages → Source to "GitHub Actions".

To run the released image:

bash
# CLIdocker run --rm -v rw-data:/data ghcr.io/sebs/rewelo:latest project list
# MCP serverdocker run --rm -i -v rw-data:/data ghcr.io/sebs/rewelo:latest serve

Use a named volume (rw-data) or a bind mount to persist the database across container restarts. Without one, the database stays inside the container and is lost when it is removed (with --read-only it can't be written at all); rw warns about this on stderr.

MCP Server

The CLI doubles as an MCP server over stdio, letting AI assistants manage projects, tickets, tags, and calculations directly. See mcp.md for client configuration, available tools, and troubleshooting.

Examples

See examples.md for step-by-step scenarios, each as a copy-pasteable Claude Code prompt (MCP) and as rw commands with their output: CSV import, dependency mapping, custom weights, quick wins, tracking work, scripting, and sharing and backing up a backlog.

來源:Readme.md,提交 260917b

工具

0
SourceWeft 尚未在沙箱中啟動過該軟體包,因此其工具未知。

版本歷史

2
  1. v0.7.3最新Sep 26, 2026
  2. v0.7.1Sep 25, 2026

評分

已滿足適用檢查項的 50%

該評分衡量此收錄條目的完整程度,依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價,也不表示伺服器安全或目前可用:可連通性是帶日期的檢查結果,評分也不包含安裝量或使用者評價。

  • 近 30 天內可連通25 分

    請保持端點能夠回應 MCP 握手;SourceWeft 每 30 天探測一次。

  • 已宣告認證方式10 分

    請在註冊中心條目中宣告憑證請求標頭,或保持端點能夠回應探測,以便判定其認證方式。

  • 已列出工具15 分

    請讓伺服器在無憑證時回應 tools/list,或發布可在沙箱中無憑證啟動的軟體包。

  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分