tracking-progress
/tracking-progress
扫描当前项目 `.scratch` 目录下的 spec、票面和 checkpoint 清单,生成一张任务进度网页。用户想知道手上有哪些活、做到哪了、现在能开工的是哪张票,或说「进度」「看一下状态」「还剩什么」时使用。
- Author
- lazygophers ↗
- Repository
- ccplugin ↗
- Licence
- AGPL-3.0
- Last repository change
- 27.10.2026
Assessment and security
- Files in this package
- 4
- Scripts
- 3
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 27.10.2026, without references to retired models or APIs.
AGPL-3.0, copyleft: przeróbki udostępniane innym muszą mieć tę samą licencję.
- 3 scripts in this package. Review them before installing, as Claude may execute them.
Scripts: scripts/mermaid.flow.min.js, scripts/test_tracking_progress.py, scripts/tracking_progress.py
What it does and when to use it
扫描当前项目 `.scratch` 目录下的 spec、票面和 checkpoint 清单,生成一张任务进度网页。用户想知道手上有哪些活、做到哪了、现在能开工的是哪张票,或说「进度」「看一下状态」「还剩什么」时使用。
Original author description
扫描当前项目 `.scratch` 目录下的 spec、票面和 checkpoint 清单,生成一张任务进度网页。用户想知道手上有哪些活、做到哪了、现在能开工的是哪张票,或说「进度」「看一下状态」「还剩什么」时使用。
How it works
- What it recognises
- What the page says (and how to read odd bits)
SKILL.md preview
--- name: tracking-progress description: 扫描当前项目 `.scratch` 目录下的 spec、票面和 checkpoint 清单,生成一张任务进度网页。用户想知道手上有哪些活、做到哪了、现在能开工的是哪张票,或说「进度」「看一下状态」「还剩什么」时使用。 --- Run the script. It does the whole job — scan, analyse, render. ```bash python3 skills/project/tracking-progress/scripts/tracking_progress.py ``` It writes `.scratch/progress.html` and prints the path. Single self-contained file: no network, no dependencies, opens from `file://` (a flowchart-only mermaid build, `scripts/mermaid.flow.min.js`, is inlined — the page weighs about 1 MB). The page is the entry directory to `.scratch/`: hash-routed views (`#/overview`, `#/now`, `#/specs`, `#/spec-<名>`, `#/memory`, `#/research`, `#/unknown`) driven by a sidebar (工作台 / 功能台 per spec / 档案库) plus top tabs — each tab/menu item is its own page, browser back/forward works. Renders light (晨雾) by default; the 深色 button (top-right) switches to dark and the choice is remembered via localStorage. Pass a repo root as the first argument to scan a different project. The page includes per-spec dependency diagrams (mermaid, colours follow the theme), and every spec name / ticket id has a copy-icon button next to it — ticket ids copy as `spec#id`. There is no 未能识别 bucket: a top-level `.scratch/<name>.md` that parses to no tasks becomes a **single-file feature** (own view, `单文件` badge); non-task files inside a feature dir (`design.md`, `brief.md`, …) become its 附件 chips, `design-demos/` (or `demos/`, `设计样例/`) is semantically recognised as a **设计样例** section: every file in it gets a gallery card in the feature view (same-stem `.png` screenshots become card thumbnails), not a generic asset chip. Every html file in a 设计样例 dir renders a **live iframe preview** as its card face (lazy-loaded, scaled to the card). Other common asset dirs (non-issues/impl/research) are listed as **expandable** entries (`<details>` open by default; subdirectories nest recursively, one `<details>` per level showing its recursive file count, files as leaves), and every remaining non-md file in the feature dir (html/png/csv screenshots etc., system junk like `.DS_Store` excluded) shows as a chip in the bottom 附件 filebox — nothing invisible. Files already listed inside an asset-dir tree are **not repeated** as filebox chips (dedup by path; only top-level loose files stay outside). In the 可开工/停滞/最近 tables the first column is the **feature** (`.scratch/<feature>/` unit — spec.md and wayfinder map are both kinds of content inside it), and the feature name links to its detail view `#/spec-<名>`. The `wayfinder 地图` / `spec` / `单文件` badges next to it are **plain labels, not clickable**; the docs live in a docbar row right under the spec head, labelled with the md's extracted h1 title (`标题(文件名)`; the same extraction now labels every md/html filename on the page — filebox chips, asset-tree leaves and demo cards; html takes `<head><title>` first, then the first `<h1>`, falling back to the plain filename when neither exists). Clicking a docbar / filebox / ticket-title link opens a **centered in-page modal** (`#pvWrap`): md renders as formatted markdown, csv/tsv as a table, json pretty-printed with basic highlighting, yaml with basic highlighting — parse failures fall back to raw text; html design docs render in a sandboxed `iframe` (`srcdoc`); images of any known extension (png/jpg/gif/svg/webp/bmp/ico/avif) inline as base64 data URIs; any other text file (utf-8 decodable, no NUL byte in the first 8 KB) shows its raw text in a monospace pre with horizontal scroll; undecodable binary files are not in the modal channel. Video attachments (mp4/m4v/mov/webm) and PDFs open in the same modal without consuming any budget: video via a native `<video controls>` element bound to the file's relative path (if the browser cannot decode it, the modal falls back to a `↗` new-tab link), PDF via an `iframe` rendered by the browser's built-in reader. Office files (docx/xlsx/pptx) and other binaries stay as `↗` new-tab links. The 2 MB inline budget covers md/text; html docs and images carry their own separate budgets (0.6 MB / 1.4 MB of file bytes, filled smallest-file-first so one big file cannot starve the rest). Files over any budget are not in the modal channel: they open in a real new tab, marked with `↗`. The top search box filters table rows as you type (⌘K focuses it). Then: 1. Open the HTML: `open .scratch/progress.html`. 2. Report the headline (the page leads with a one-sentence conclusion) and the 现在能开工的 table. Link the HTML as `[任务进度](.scratch/progress.html)`. 3. Open the page's 未能识别的文件 section if it is non-empty. For each file, decide whether it actually holds tasks. If any does, the script is missing a convention — ask the user whether to file an issue at `.scratch/tracking-progress/issues/` describing the unrecognised format, then add the rule to the script. (The script prints only the output path; the count, if any, lives on the page.) ## What it recognises | Where tasks live | How status is read | | --- | --- | | `<feature>/issues/NN-slug.md`, `<feature>/impl/NN-slug.md`, `<feature>/NN-slug.md` | `Status:` / `状态:` line, bare or bold | | `issues/open/` vs `issues/done/` | the directory name | | ticket file with `Blocked by:` / `Type:` but no status | open (nobody marked it done) | | `memory.md` checkpoint rows | `- [x]` done, `- [ ]` open, `- [~]` / `- [-]` in progress | | wayfinder `map.md` + its children | `Status: claimed` / `resolved`, `Blocked by: NN` | `memory/` and the top-level `research/` are never scanned for tasks — they are history, not work in flight. The 决策档案 (`#/memory`) view renders the ledger as 变更审计流: day stubs with counts only; clicking a day parses that day's text on demand from the inlined `tp-files` JSON (canonical `- 改成: … | 原本 | 我说 | 为什么 | 否掉了` lines become diff cards, everything else stays as verbatim 原文 blocks — nothing hidden, nothing dropped). Decision content is never written into the HTML at generation time, so page size and DOM stay bounded. A feature-level `<feature>/research/*.md` is listed under its feature with links (in-page modal, same `#pvWrap` channel), status-free. Status strings are matched loosely: `resolved` / `done` / `已完成` / `定稿` all count as done, `claimed` / `in-progress` / `进行中` as in progress, everything else as open. A negated status (`未完成`, `无法完成`, `not done`, `unresolved`) counts as open, not done — the negation prefix cancels the keyword hit. ## What the page says (and how to read odd bits) - **Blocking only counts ids that exist.** A `Blocked by:` referencing a number no ticket owns (often a document reference like `ADR-0046` minted into an id) is shown as a dangling ref, not treated as a blocker. - **互相阻塞的票** (A blocks B, B blocks A) are drawn with dashed amber edges in the wave diagram — someone has to break the loop by hand. - **未建票 vs 票已建齐**: a map/checklist row with no ticket file behind it is marked 未建 (badge on the spec head, mark on the row, dashed node in the graph, count in the per-spec table). A spec where every row has a ticket file gets the 票已建齐 badge. - **口径写在页面上**: progress = done ÷ total, blockers in the denominator; one ticket = one ticket file or one `memory.md` checklist row, so "done" can include process notes like 「写 spec.md」. The page states this instead of hiding it.
Files in this package (4)
- SKILL.md 8 KB
- scripts/mermaid.flow.min.js 999 KB
- scripts/test_tracking_progress.py 38 KB
- scripts/tracking_progress.py 124 KB
Versions
The database has no other skills named tracking-progress.
Database snapshot: 13.11.2026 · package version 8547562d1a15 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100request-tracking
Request tracking knowledge base. Use when: querying request status, managing document references, tracking progress. Not for: creating requests (use create-request), tech specs (use tech-spec). Output: status report + progress tracking. - 100progress-tracking
Validates feature plans, tracks implementation progress, and verifies code-level requirements traceability. Use when validating a feature plan, tracking its progress, or checking requirements against code. - 91create-request
Create, update, or scan per-task request tickets for progress tracking. These are date-prefixed non-lifecycle docs under requests/, NOT feature-level requirements (use /req-analyze for those). Use when: tracking task progress, updating completion status, scanning incomplete… - 100checkpoint
Save progress checkpoint — git commit WIP and update Linear issue with progress comment. Use when the user says 'checkpoint', 'save progress', 'wip commit', or invokes /checkpoint. - 90working-on
Set a tracking document as the source of truth for the current feature or task. Use when starting work on a feature, bug fix, or multi-step task that benefits from a persistent record of decisions, discoveries, and progress. Keeps the document updated as work proceeds. - 100progress-tracking
Visual progress reporting — progress bars, done / in-progress / remaining checklists, and completion estimates so a non-technical user always knows where the build stands. Transparency without overwhelm. Use during multi-step builds and whenever the user asks how far along the work is.