[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M755E0WG3N8XGTS291CWECCE":3,"skill-same:01M755E0WG3N8XGTS291CWECCE":58},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":31,"install":32,"resources":45},"01M755E0WG3N8XGTS291CWECCE","change-log",null,"Map a task's changes as an editable draw that renders an interactive change-map on Save — nodes are functional CHANGES (a capability that moved), arrows are dependencies, and clicking a change deep-dives into its file \u002F function \u002F test detail in a side inspector. Use when someone asks \"what changed\", \"map this change\", \"diagram this PR\u002Fbranch\", \"show how the changes connect\", or as the closing step of a task once code has landed. Reads the diff + the task's manifest (the *why*) + Hub's MCP task context, then writes ONE `.excalidraw` into `tasks\u002F\u003Cslug>\u002Fchange-log\u002F`; the user edits it and Saving renders the sibling HTML. The agent does the reading — Hub has no model. `\u002Fchange-log \u003Ctask-slug> [--since \u003Cref>] [--doc]`.","auth-02\u002Fhub","auth-02","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F146264364?v=4","01M755DW6DVATA25MM8323VK52","Atharva Shinde","01M755DW7CDZNCA30JT6VP7MQ5",3,"MIT","https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub\u002Fblob\u002Fmain\u002Fhubspace\u002Fplugin\u002Fhub-agent\u002Fskills\u002Fchange-log\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: change-log\ndescription: Map a task's changes as an editable draw that renders an interactive change-map on Save — nodes are functional CHANGES (a capability that moved), arrows are dependencies, and clicking a change deep-dives into its file \u002F function \u002F test detail in a side inspector. Use when someone asks \"what changed\", \"map this change\", \"diagram this PR\u002Fbranch\", \"show how the changes connect\", or as the closing step of a task once code has landed. Reads the diff + the task's manifest (the *why*) + Hub's MCP task context, then writes ONE `.excalidraw` into `tasks\u002F\u003Cslug>\u002Fchange-log\u002F`; the user edits it and Saving renders the sibling HTML. The agent does the reading — Hub has no model. `\u002Fchange-log \u003Ctask-slug> [--since \u003Cref>] [--doc]`.\n---\n\n# change-log — the agent reads the diff, Hub renders an explorable map\n\nThe point of a change-log here is **a picture you can interrogate, not prose**:\neach **node is a functional change** — a *capability that moved* (\"delete a\ncomment\", \"publish reach\"), NOT a file — and each **arrow is a dependency**\n(\"calls\" \u002F \"feeds\" \u002F \"renames\"). A reviewer glances at the flow to see the shape\nof the change, then **clicks any node to deep-dive** into that change's files,\nfunctions\u002Fsymbols, and tests in a side inspector. (Modeled on workflow-map tools\nlike HQFlow: the agent authors the graph, the page renders it + the inspector.)\n\nReading a diff and deciding what the *functional changes* are is the one job Hub\ncannot do. Hub has no model, no network, no API key — that is the whole point. So\nchange-log is **a skill your agent runs**: you read the diff, gather the task's\ncontext, reason out the change-graph (each node carrying its own deep-dive\ndetail), and drop ONE self-contained HTML page into the task's `artifacts\u002F`. From\nHub's side nothing is special — a file appeared, the watcher indexed it, it has\nlineage, it opens in the workspace. *Hub stays a consumer even when the thing it\nconsumes is intelligence.*\n\n```\n\u002Fchange-log \u003Ctask-slug> [--since \u003Cref>] [--doc]\n```\n\n- `\u003Ctask-slug>` — the task whose work you are mapping (its folder under `tasks\u002F`).\n- `--since \u003Cref>` — git ref to diff from (branch, tag, or commit). Default: the\n  merge-base with `main` (`git merge-base main HEAD`), i.e. everything on this branch.\n- `--doc` — ALSO write the prose HTML write-up (`templates\u002Fchange-log.html`) into\n  `artifacts\u002F` when someone wants a linear narrative.\n\nThe **editable draw → interactive map on Save** is the primary deliverable; the\ndraw is the single source of truth (the map is rendered from it). `--doc` is an\noptional narrative companion.\n\n## Step 1 — read three sources (this is the work)\n\nYou are the only thing here that can read a diff. Gather:\n\n1. **What changed** — the diff:\n   ```bash\n   git diff \u003Csince>..HEAD            # \u003Csince> defaults to $(git merge-base main HEAD)\n   git diff --stat \u003Csince>..HEAD     # the files-touched table + rough sizes\n   ```\n2. **Why it changed** — the task's manifest, `tasks\u002F\u003Cslug>\u002Fmanifest.md`. The\n   *Decisions* and *Plan* sections are where the one-line \"why\" for each change\n   comes from. A diff says *what*; the manifest says *why*. Use both.\n3. **The task's shape** — Hub's MCP read tools (from `hub mcp serve`):\n   - `get_task(slug)` → status, plan, decisions, lineage counts.\n   - `trace(path)` → the lineage around a file.\n\n   **If MCP is not running, fall back** to reading the files directly:\n   ```bash\n   cat tasks\u002F\u003Cslug>\u002Fmanifest.md\n   hub timeline \u003Cslug> --json        # the task's evolution spine\n   ```\n\nNever invent a \"why\". If the manifest does not explain a change, leave that\nnode's note blank rather than guessing.\n\n## Step 2 — reason out the change-graph (nodes + edges)\n\nThis is the judgement call, and it is yours. Collapse the raw diff into a handful\nof **functional changes** — NOT one node per file. A node is a *capability that\nmoved*: \"delete a comment\", \"publish reach\", \"settings panel\". Aim for **5–12\nnodes**; if you have more, you are mapping code, not the change. Then draw the\n**dependency arrows** between them (\"A calls B\", \"B feeds A\") — the path a\nreviewer traces to understand the PR.\n\nThe file\u002Ffunction\u002Ftest detail does **not** go in the node — it goes in the\nnode's **`details`**, surfaced when the reviewer clicks to deep-dive. Node\n(change-oriented on top, file-oriented underneath):\n\n```json\n{ \"id\": \"endpoint\",\n  \"kind\": \"task\",                       \u002F\u002F category → accent colour (see palette)\n  \"title\": \"Delete a comment\",          \u002F\u002F the CHANGE, a short noun phrase\n  \"verb\": \"new\",                        \u002F\u002F new | changed | rewritten | removed\n  \"summary\": \"Remove one comment by id, mirroring the add path.\",\n  \"files\":     [{\"path\":\"hubspace\u002Fcli\u002Fserver.py\",\"change\":\"+handler\"}],\n  \"functions\": [{\"symbol\":\"_note_delete()\",\"note\":\"guards + rebuild\"}],\n  \"tests\":     [\"TestNoteDeleteEndpoint\"],\n  \"note\": \"mirrors \u002F_note\" }\n```\n\n`title` + `summary` are what shows on the card; `files` \u002F `functions` \u002F `tests`\n\u002F `note` are the **deep-dive**, shown in the inspector on click. Draw them from\nthe diff (files\u002Fsymbols) and the manifest (the *why*); leave a list empty rather\nthan inventing. Each edge is `{ \"from\", \"to\", \"rel\"? }` — an arrow *from* the\ndependent *to* what it depends on; `rel` is a short relationship word drawn on\nthe arrow.\n\n**Kind → accent colour** (reuse Hub's palette; any other kind renders neutral):\n`task` oxblood · `artifact` violet · `script` slate · `doc` deep-sea ·\n`data` teal · `run` green · `draw` amber · `note` rust · `prompt` gold.\nPick the kind that fits the change (endpoint\u002Ffeature → `task`, helper\u002Fmodule →\n`script`, UI\u002Fasset\u002Fpage → `artifact`, spec\u002Fdocs → `doc`).\n\n## Step 3 — write the editable draw; Save renders the interactive map\n\n**View-first, edit-on-demand.** The default surface a reader sees is the\ninteractive **HTML** map (click-to-deep-dive: numbered accent cards, labelled\narrows, no grid, Purpose · Files · Functions · Tests · Note inspector). It\ncarries an **\"✎ Edit\" button** → the editable Excalidraw **canvas**; the canvas\ncarries a **\"▶ interactive version\"** pill back. Editing then Saving in the\ncanvas **re-renders the HTML** (Hub reconstructs the graph from the scene — label\n+ position edits flow through). So the loop is *view → edit → save → view*.\n\nYou author BOTH files in the task's **`change-log\u002F`** folder: the `.excalidraw`\n(source of truth; `changelog.to_scene` embeds your whole change-graph — each\nnode's deep-dive detail — in element `customData`) and a pre-rendered `.html`\n(the default view) so the reader lands on the map immediately.\n\n```\ntasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.html          ← default view (the reader opens this)\ntasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.excalidraw    ← editable source (reached via ✎ Edit)\n```\n\n```bash\npython3 - \u003C\u003C'PY'\nimport json\nfrom hubspace.core import changelog\nmeta = {\n  \"title\": \"Change-log — delete a comment\", \"slug\": \"\u003Cslug>\",\n  \"subtitle\": \"\u003Cshort-since>..\u003Cshort-head> · 2 changes\",\n  # provenance — surfaces as the map's \"written by …\" line + seeds \"ask again\":\n  \"generated_by\": \"claude ▸ skill:change-log\",\n  \"commit_range\": \"\u003Cshort-since>..\u003Cshort-head>\",\n  \"written_at\": \"\u003CISO-8601 timestamp>\",\n}\nnodes = [\n  {\"id\":\"endpoint\",\"kind\":\"task\",\"title\":\"Delete a comment\",\"verb\":\"new\",\n   \"summary\":\"Remove one comment by id, mirroring the add path.\",\n   \"files\":[{\"path\":\"hubspace\u002Fcli\u002Fserver.py\",\"change\":\"+handler\"}],\n   \"functions\":[{\"symbol\":\"_note_delete()\",\"note\":\"guards + rebuild\"}],\n   \"tests\":[\"TestNoteDeleteEndpoint\"],\"note\":\"mirrors \u002F_note\"},\n  {\"id\":\"store\",\"kind\":\"script\",\"title\":\"Notes store\",\"verb\":\"new\",\n   \"summary\":\"Byte-preserving line removal from notes.jsonl.\",\n   \"files\":[{\"path\":\"hubspace\u002Fcore\u002Ftasks.py\",\"change\":\"+delete_note\"}],\n   \"functions\":[{\"symbol\":\"delete_note()\",\"note\":\"idempotent\"}],\n   \"tests\":[\"TestDeleteNote\"]},\n]\nedges = [ {\"from\":\"endpoint\",\"to\":\"store\",\"rel\":\"calls\"} ]\ndraw_href = \"\u002Ftasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.excalidraw\"\nhtml_href = \"\u002Ftasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.html\"\nscene = changelog.to_scene(nodes, edges, title=meta[\"title\"], subtitle=meta[\"subtitle\"],\n    meta=meta, interactive_href=html_href)\nopen(\"tasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.excalidraw\",\"w\").write(json.dumps(scene, ensure_ascii=False))\n# Pre-render the default HTML view; edit_href gives it the ✎ Edit → canvas button.\nfrom hubspace.core import changemap\ng = changelog.scene_to_graph(scene)\ng[\"meta\"][\"edit_href\"] = draw_href\nopen(\"tasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.html\",\"w\").write(\n    changemap.render_html(g[\"meta\"], g[\"nodes\"], g[\"edges\"], positions=g[\"positions\"]))\nPY\n```\n\nGet the short hashes with `git rev-parse --short \u003Csince>` \u002F `--short HEAD`. Keep\nthe **cards** high level (the change + summary); push files, symbols and tests\ninto each node's `details` for the deep-dive. Point the reader at the `.html`.\n\n**How Save → HTML works:** Hub watches for `.excalidraw` saves under any\n`change-log\u002F` folder; on Save it runs `changelog.scene_to_graph(scene)` →\n`changemap.render_html(...)` and rewrites the `.html` sibling (setting `edit_href`\nback to the draw automatically). Label + position edits on the canvas flow\nthrough — the model rides in the scene's `customData`, and an edited card label\nis recovered whether Excalidraw kept our tagged text, bound a new text to the\ncard, or (worst case) dropped the tag (a positional fallback reads the card's\nlargest-font text). The page carries `hub:standalone`, so Hub serves it without\ninjecting doc chrome, self-contained\u002Foffline for `hub publish`.\n\n## `--doc` — the prose companion (optional)\n\nWith `--doc`, ALSO copy `templates\u002Fchange-log.html` and fill it (Title → Summary\n→ Before → After → Files touched → Provenance), writing to\n`tasks\u002F\u003Cslug>\u002Fartifacts\u002Fchange-log-\u003CYYYY-MM-DD>-notes.html`. A linear narrative\nfor when someone wants to read rather than explore. Same self-contained\u002Foffline\nrule so `hub publish` can ship it.\n\n### Provenance — how Hub learns who made it\n\nThe interactive map (`changemap.render_html`) emits the provenance front matter\nfor you from the `meta` `generated_by` \u002F `commit_range` \u002F `written_at` \u002F `slug`\nfields (a top-of-file HTML comment). For the `--doc` template, put the same block\nat the **very top** by hand:\n\n```html\n\u003C!--\n---\ngenerated_by: \"claude ▸ skill:change-log\"\ncommit_range: \"\u003Cshort-since>..\u003Cshort-head>\"\nwritten_at: \u003CISO-8601 timestamp>\ntask: \u003Cslug>\n---\n-->\n\u003C!DOCTYPE html>\n…\n```\n\nHub reads those four fields and renders `written by \u003Cgenerated_by> · \u003Cwritten_at>\n· \u003Ccommit_range>` plus **\"Hub did not generate this file.\"**, and seeds the\ncopy-only **\"ask again\"** button (`\u002Fchange-log \u003Cslug> --since \u003Cend>`).\n\n## What this skill must NOT do\n\n- **No model \u002F network \u002F key in Hub.** You (the agent) do the reasoning — reading\n  the diff, choosing the functional changes, their dependencies, and each change's\n  file\u002Ffunction\u002Ftest detail. Hub only lays out the graph you hand it (deterministic)\n  , indexes the file, and reads front matter. Do not add generation code to Hub.\n- Map the **change**, not the code — 5–12 functional-change nodes; the file\u002Fsymbol\n  detail belongs in each node's `details` (the deep-dive), not as its own node.\n  One-node-per-file is the failure mode.\n- Write the draw into `tasks\u002F\u003Cslug>\u002Fchange-log\u002F` (Hub renders the `.html` there\n  on Save); `--doc` goes in `tasks\u002F\u003Cslug>\u002Fartifacts\u002F`.\n- Do not fabricate — if the manifest is silent on a \"why\", leave the note\u002Fsummary\n  spare rather than inventing one.\n\n## After writing\n\nMention the path. The watcher indexes it within ~3 s; the page appears under its\ntask with full lineage and opens in the workspace — click a change to deep-dive.\nTo share it, hand off to `hub publish` \u002F the `dak` skill.\n","# change-log — the agent reads the diff, Hub renders an explorable map\n\nThe point of a change-log here is **a picture you can interrogate, not prose**:\neach **node is a functional change** — a *capability that moved* (\"delete a\ncomment\", \"publish reach\"), NOT a file — and each **arrow is a dependency**\n(\"calls\" \u002F \"feeds\" \u002F \"renames\"). A reviewer glances at the flow to see the shape\nof the change, then **clicks any node to deep-dive** into that change's files,\nfunctions\u002Fsymbols, and tests in a side inspector. (Modeled on workflow-map tools\nlike HQFlow: the agent authors the graph, the page renders it + the inspector.)\n\nReading a diff and deciding what the *functional changes* are is the one job Hub\ncannot do. Hub has no model, no network, no API key — that is the whole point. So\nchange-log is **a skill your agent runs**: you read the diff, gather the task's\ncontext, reason out the change-graph (each node carrying its own deep-dive\ndetail), and drop ONE self-contained HTML page into the task's `artifacts\u002F`. From\nHub's side nothing is special — a file appeared, the watcher indexed it, it has\nlineage, it opens in the workspace. *Hub stays a consumer even when the thing it\nconsumes is intelligence.*\n\n```\n\u002Fchange-log \u003Ctask-slug> [--since \u003Cref>] [--doc]\n```\n\n- `\u003Ctask-slug>` — the task whose work you are mapping (its folder under `tasks\u002F`).\n- `--since \u003Cref>` — git ref to diff from (branch, tag, or commit). Default: the\n  merge-base with `main` (`git merge-base main HEAD`), i.e. everything on this branch.\n- `--doc` — ALSO write the prose HTML write-up (`templates\u002Fchange-log.html`) into\n  `artifacts\u002F` when someone wants a linear narrative.\n\nThe **editable draw → interactive map on Save** is the primary deliverable; the\ndraw is the single source of truth (the map is rendered from it). `--doc` is an\noptional narrative companion.\n\n## Step 1 — read three sources (this is the work)\n\nYou are the only thing here that can read a diff. Gather:\n\n1. **What changed** — the diff:\n   ```bash\n   git diff \u003Csince>..HEAD            # \u003Csince> defaults to $(git merge-base main HEAD)\n   git diff --stat \u003Csince>..HEAD     # the files-touched table + rough sizes\n   ```\n2. **Why it changed** — the task's manifest, `tasks\u002F\u003Cslug>\u002Fmanifest.md`. The\n   *Decisions* and *Plan* sections are where the one-line \"why\" for each change\n   comes from. A diff says *what*; the manifest says *why*. Use both.\n3. **The task's shape** — Hub's MCP read tools (from `hub mcp serve`):\n   - `get_task(slug)` → status, plan, decisions, lineage counts.\n   - `trace(path)` → the lineage around a file.\n\n   **If MCP is not running, fall back** to reading the files directly:\n   ```bash\n   cat tasks\u002F\u003Cslug>\u002Fmanifest.md\n   hub timeline \u003Cslug> --json        # the task's evolution spine\n   ```\n\nNever invent a \"why\". If the manifest does not explain a change, leave that\nnode's note blank rather than guessing.\n\n## Step 2 — reason out the change-graph (nodes + edges)\n\nThis is the judgement call, and it is yours. Collapse the raw diff into a handful\nof **functional changes** — NOT one node per file. A node is a *capability that\nmoved*: \"delete a comment\", \"publish reach\", \"settings panel\". Aim for **5–12\nnodes**; if you have more, you are mapping code, not the change. Then draw the\n**dependency arrows** between them (\"A calls B\", \"B feeds A\") — the path a\nreviewer traces to understand the PR.\n\nThe file\u002Ffunction\u002Ftest detail does **not** go in the node — it goes in the\nnode's **`details`**, surfaced when the reviewer clicks to deep-dive. Node\n(change-oriented on top, file-oriented underneath):\n\n```json\n{ \"id\": \"endpoint\",\n  \"kind\": \"task\",                       \u002F\u002F category → accent colour (see palette)\n  \"title\": \"Delete a comment\",          \u002F\u002F the CHANGE, a short noun phrase\n  \"verb\": \"new\",                        \u002F\u002F new | changed | rewritten | removed\n  \"summary\": \"Remove one comment by id, mirroring the add path.\",\n  \"files\":     [{\"path\":\"hubspace\u002Fcli\u002Fserver.py\",\"change\":\"+handler\"}],\n  \"functions\": [{\"symbol\":\"_note_delete()\",\"note\":\"guards + rebuild\"}],\n  \"tests\":     [\"TestNoteDeleteEndpoint\"],\n  \"note\": \"mirrors \u002F_note\" }\n```\n\n`title` + `summary` are what shows on the card; `files` \u002F `functions` \u002F `tests`\n\u002F `note` are the **deep-dive**, shown in the inspector on click. Draw them from\nthe diff (files\u002Fsymbols) and the manifest (the *why*); leave a list empty rather\nthan inventing. Each edge is `{ \"from\", \"to\", \"rel\"? }` — an arrow *from* the\ndependent *to* what it depends on; `rel` is a short relationship word drawn on\nthe arrow.\n\n**Kind → accent colour** (reuse Hub's palette; any other kind renders neutral):\n`task` oxblood · `artifact` violet · `script` slate · `doc` deep-sea ·\n`data` teal · `run` green · `draw` amber · `note` rust · `prompt` gold.\nPick the kind that fits the change (endpoint\u002Ffeature → `task`, helper\u002Fmodule →\n`script`, UI\u002Fasset\u002Fpage → `artifact`, spec\u002Fdocs → `doc`).\n\n## Step 3 — write the editable draw; Save renders the interactive map\n\n**View-first, edit-on-demand.** The default surface a reader sees is the\ninteractive **HTML** map (click-to-deep-dive: numbered accent cards, labelled\narrows, no grid, Purpose · Files · Functions · Tests · Note inspector). It\ncarries an **\"✎ Edit\" button** → the editable Excalidraw **canvas**; the canvas\ncarries a **\"▶ interactive version\"** pill back. Editing then Saving in the\ncanvas **re-renders the HTML** (Hub reconstructs the graph from the scene — label\n+ position edits flow through). So the loop is *view → edit → save → view*.\n\nYou author BOTH files in the task's **`change-log\u002F`** folder: the `.excalidraw`\n(source of truth; `changelog.to_scene` embeds your whole change-graph — each\nnode's deep-dive detail — in element `customData`) and a pre-rendered `.html`\n(the default view) so the reader lands on the map immediately.\n\n```\ntasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.html          ← default view (the reader opens this)\ntasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.excalidraw    ← editable source (reached via ✎ Edit)\n```\n\n```bash\npython3 - \u003C\u003C'PY'\nimport json\nfrom hubspace.core import changelog\nmeta = {\n  \"title\": \"Change-log — delete a comment\", \"slug\": \"\u003Cslug>\",\n  \"subtitle\": \"\u003Cshort-since>..\u003Cshort-head> · 2 changes\",\n  # provenance — surfaces as the map's \"written by …\" line + seeds \"ask again\":\n  \"generated_by\": \"claude ▸ skill:change-log\",\n  \"commit_range\": \"\u003Cshort-since>..\u003Cshort-head>\",\n  \"written_at\": \"\u003CISO-8601 timestamp>\",\n}\nnodes = [\n  {\"id\":\"endpoint\",\"kind\":\"task\",\"title\":\"Delete a comment\",\"verb\":\"new\",\n   \"summary\":\"Remove one comment by id, mirroring the add path.\",\n   \"files\":[{\"path\":\"hubspace\u002Fcli\u002Fserver.py\",\"change\":\"+handler\"}],\n   \"functions\":[{\"symbol\":\"_note_delete()\",\"note\":\"guards + rebuild\"}],\n   \"tests\":[\"TestNoteDeleteEndpoint\"],\"note\":\"mirrors \u002F_note\"},\n  {\"id\":\"store\",\"kind\":\"script\",\"title\":\"Notes store\",\"verb\":\"new\",\n   \"summary\":\"Byte-preserving line removal from notes.jsonl.\",\n   \"files\":[{\"path\":\"hubspace\u002Fcore\u002Ftasks.py\",\"change\":\"+delete_note\"}],\n   \"functions\":[{\"symbol\":\"delete_note()\",\"note\":\"idempotent\"}],\n   \"tests\":[\"TestDeleteNote\"]},\n]\nedges = [ {\"from\":\"endpoint\",\"to\":\"store\",\"rel\":\"calls\"} ]\ndraw_href = \"\u002Ftasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.excalidraw\"\nhtml_href = \"\u002Ftasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.html\"\nscene = changelog.to_scene(nodes, edges, title=meta[\"title\"], subtitle=meta[\"subtitle\"],\n    meta=meta, interactive_href=html_href)\nopen(\"tasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.excalidraw\",\"w\").write(json.dumps(scene, ensure_ascii=False))\n# Pre-render the default HTML view; edit_href gives it the ✎ Edit → canvas button.\nfrom hubspace.core import changemap\ng = changelog.scene_to_graph(scene)\ng[\"meta\"][\"edit_href\"] = draw_href\nopen(\"tasks\u002F\u003Cslug>\u002Fchange-log\u002F\u003Cname>.html\",\"w\").write(\n    changemap.render_html(g[\"meta\"], g[\"nodes\"], g[\"edges\"], positions=g[\"positions\"]))\nPY\n```\n\nGet the short hashes with `git rev-parse --short \u003Csince>` \u002F `--short HEAD`. Keep\nthe **cards** high level (the change + summary); push files, symbols and tests\ninto each node's `details` for the deep-dive. Point the reader at the `.html`.\n\n**How Save → HTML works:** Hub watches for `.excalidraw` saves under any\n`change-log\u002F` folder; on Save it runs `changelog.scene_to_graph(scene)` →\n`changemap.render_html(...)` and rewrites the `.html` sibling (setting `edit_href`\nback to the draw automatically). Label + position edits on the canvas flow\nthrough — the model rides in the scene's `customData`, and an edited card label\nis recovered whether Excalidraw kept our tagged text, bound a new text to the\ncard, or (worst case) dropped the tag (a positional fallback reads the card's\nlargest-font text). The page carries `hub:standalone`, so Hub serves it without\ninjecting doc chrome, self-contained\u002Foffline for `hub publish`.\n\n## `--doc` — the prose companion (optional)\n\nWith `--doc`, ALSO copy `templates\u002Fchange-log.html` and fill it (Title → Summary\n→ Before → After → Files touched → Provenance), writing to\n`tasks\u002F\u003Cslug>\u002Fartifacts\u002Fchange-log-\u003CYYYY-MM-DD>-notes.html`. A linear narrative\nfor when someone wants to read rather than explore. Same self-contained\u002Foffline\nrule so `hub publish` can ship it.\n\n### Provenance — how Hub learns who made it\n\nThe interactive map (`changemap.render_html`) emits the provenance front matter\nfor you from the `meta` `generated_by` \u002F `commit_range` \u002F `written_at` \u002F `slug`\nfields (a top-of-file HTML comment). For the `--doc` template, put the same block\nat the **very top** by hand:\n\n```html\n\u003C!--\n---\ngenerated_by: \"claude ▸ skill:change-log\"\ncommit_range: \"\u003Cshort-since>..\u003Cshort-head>\"\nwritten_at: \u003CISO-8601 timestamp>\ntask: \u003Cslug>\n---\n-->\n\u003C!DOCTYPE html>\n…\n```\n\nHub reads those four fields and renders `written by \u003Cgenerated_by> · \u003Cwritten_at>\n· \u003Ccommit_range>` plus **\"Hub did not generate this file.\"**, and seeds the\ncopy-only **\"ask again\"** button (`\u002Fchange-log \u003Cslug> --since \u003Cend>`).\n\n## What this skill must NOT do\n\n- **No model \u002F network \u002F key in Hub.** You (the agent) do the reasoning — reading\n  the diff, choosing the functional changes, their dependencies, and each change's\n  file\u002Ffunction\u002Ftest detail. Hub only lays out the graph you hand it (deterministic)\n  , indexes the file, and reads front matter. Do not add generation code to Hub.\n- Map the **change**, not the code — 5–12 functional-change nodes; the file\u002Fsymbol\n  detail belongs in each node's `details` (the deep-dive), not as its own node.\n  One-node-per-file is the failure mode.\n- Write the draw into `tasks\u002F\u003Cslug>\u002Fchange-log\u002F` (Hub renders the `.html` there\n  on Save); `--doc` goes in `tasks\u002F\u003Cslug>\u002Fartifacts\u002F`.\n- Do not fabricate — if the manifest is silent on a \"why\", leave the note\u002Fsummary\n  spare rather than inventing one.\n\n## After writing\n\nMention the path. The watcher indexes it within ~3 s; the page appears under its\ntask with full lineage and opens in the workspace — click a change to deep-dive.\nTo share it, hand off to `hub publish` \u002F the `dak` skill.\n",{"name":5,"description":7},"main","dc89ede1053a613e8f3cbe025b605a2486ffb8c1","hubspace\u002Fplugin\u002Fhub-agent\u002Fskills\u002Fchange-log",0,"https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub","https:\u002F\u002Fauth-02.github.io\u002Fhub\u002F","https:\u002F\u002Fgithub.com\u002Fauth-02","2026-08-12 15:23:18",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add auth-02\u002Fhub --skill change-log",{"tool":38,"command":39},"bunx","bunx skills add auth-02\u002Fhub --skill change-log",{"tool":41,"command":42},"pnpm","pnpm dlx skills add auth-02\u002Fhub --skill change-log","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub\u002Fblob\u002Fmain\u002Fhubspace\u002Fplugin\u002Fhub-agent\u002Fskills\u002Fchange-log\u002FSKILL.md",true,[46,52],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"templates\u002Fchange-log.excalidraw","OTHER","其他",27746,"https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub\u002Fblob\u002Fmain\u002Fhubspace\u002Fplugin\u002Fhub-agent\u002Fskills\u002Fchange-log\u002Ftemplates\u002Fchange-log.excalidraw",{"path":53,"resourceType":54,"resourceTypeDesc":55,"sizeBytes":56,"sourceHtmlUrl":57},"templates\u002Fchange-log.html","REFERENCE","参考文档",4731,"https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub\u002Fblob\u002Fmain\u002Fhubspace\u002Fplugin\u002Fhub-agent\u002Fskills\u002Fchange-log\u002Ftemplates\u002Fchange-log.html",[59,66,73,80,86,92],{"skillNo":60,"name":61,"nameZh":6,"description":62,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":19},"01M755ES8KW0306XX82VJHMJNJ","stacked","Decompose LARGE, complex, or risky changes into a stack of small, dependent, independently reviewable units. Use only when the change is genuinely big: multiple architectural layers with functionally distinct concerns (models + storage + logic + API + UI), or a diff too large to review as one unit. Trigger on \"stacked\" \u002F \"stacked PRs\" \u002F \"stack this\" \u002F how to break down a large change — or when a \"build X\" \u002F \"refactor W\" request is clearly multi-layer and large. Do NOT use for small or moderate changes: bug fixes, tweaks, config, single-file or single-layer changes belong on ONE branch as ONE change — a stack of tiny branches adds overhead without value.","https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub\u002Fblob\u002Fmain\u002Fhubspace\u002Fplugin\u002Fhub-agent\u002Fskills\u002Fstacked\u002FSKILL.md",[],[],{"skillNo":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":19},"01M755EKGJEC3AXNZJ1T1C6582","manifest","Use when starting a non-trivial feature, migration, refactor, or multi-step task in ANY repo. Creates and continuously maintains a living markdown manifest at `\u003Crepo-root>\u002Ftasks\u002F\u003Cslug>\u002Fmanifest.md` (inside the repo being worked on — NEVER the skill dir or ~\u002F) capturing the problem, decisions, exploration results, plan, open questions, test + rollout plans, and final review. Trigger on \"build X\", \"add X\", \"migrate X\", \"refactor X\", tasks needing 3+ steps, or whenever a plan-before-code loop would help. NOT for single-line fixes, obvious bugs, or one-shot questions.","https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub\u002Fblob\u002Fmain\u002Fhubspace\u002Fplugin\u002Fhub-agent\u002Fskills\u002Fmanifest\u002FSKILL.md",[],[],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":19},"01M755EBMJGHXN2WGF9T0RG755","kagaz","Kagaz (कागज़) — a personal design system for all frontend and document work — editorial-technical-archival, serif display + mono metadata on warm paper tones with a single rust accent. Use for ANY frontend, UI, web page, landing page, dashboard, artifact, component, mockup, poster, slide deck, report, or PDF\u002Fdocument design, for HTML→PDF conversion, and for any \"make this look good\" \u002F \"style this\" \u002F \"make slides\" \u002F \"turn this into a PDF\" request — even when no design preference is mentioned. Overrides generic AI defaults (Inter-on-white, purple gradients, cool grays). Ships an HTML-to-PDF converter with visual QA and on-brand SVG charts.","https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub\u002Fblob\u002Fmain\u002Fhubspace\u002Fplugin\u002Fhub-agent\u002Fskills\u002Fkagaz\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":19},"01M755E9RR0Q8FNNXS5XKY88AZ","dak","https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub\u002Fblob\u002Fmain\u002Fhubspace\u002Fplugin\u002Fhub-agent\u002Fskills\u002Fdak\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":19},"01M755DZDR0YRNBDE7RXEFDAXA","auth","https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub\u002Fblob\u002Fmain\u002Fexample\u002Fweb-dashboard\u002Fapp\u002Fskills\u002Fauth\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":19},"01M755DY7A7MEMV28B9DYWXAJZ","rate_limiting","https:\u002F\u002Fgithub.com\u002Fauth-02\u002Fhub\u002Fblob\u002Fmain\u002Fexample\u002Facme-api\u002Fapp\u002Fskills\u002Frate_limiting\u002FSKILL.md",[],[]]