[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EDAN331EYXY6MJPE542NHT":3,"skill-same:01M7EDAN331EYXY6MJPE542NHT":49},{"skillNo":4,"name":5,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":20,"frontMatter":21,"refName":22,"commitSha":23,"skillDirectory":24,"forksCount":13,"repositoryHtmlUrl":25,"repositoryHomepageUrl":6,"ownerHtmlUrl":26,"ownerBlogUrl":6,"ownerDescription":27,"updatedAt":28,"install":29,"resources":42},"01M7EDAN331EYXY6MJPE542NHT","no-outlinks-audit",null,"jonahthan433\u002FCortexOS","jonahthan433","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F211465495?v=4","01M7ECH9HPHGA89DCXHHJBTS3C","jonah.ai","01M7ECH9JCEM45DZMWPT7148S1",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fjonahthan433\u002FCortexOS\u002Fblob\u002Fmain\u002Fskills\u002Fimported\u002Fproductivity\u002Fno-outlinks-audit\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"# no-outlinks-audit\n\n# SKILL: Dead-End Pages — No Outgoing Internal Links Audit\n\n## DESCRIPTION\n\nFinds every blog\u002Fcontent page on a site that has **zero outgoing links to any URL on\nthe same domain** in its body content. These pages absorb link equity but pass none\non — they are structural dead-ends that silently suppress topical authority and hurt\ncrawl efficiency across the site.\n\nFor each dead-end page the skill generates 3 suggestions: pages on the same site that\nthe dead-end page SHOULD link out to, including the exact anchor text (the target\npage's top keyword), where in the dead-end page to place the link, and a ready-to-paste\ncontext copy sentence with the link already embedded.\n\nThis skill is the structural inverse of the Orphan Pages skill:\n- **Orphan Pages**:   pages with no *incoming* internal links (nobody links TO them)\n- **Dead-End Pages**: pages with no *outgoing* internal links (they link out to nobody)\n\n---\n\n## TOOLS\n\n- Step 1 uses **curl via bash_tool** for sitemap discovery\n- Step 2 uses **curl via bash_tool** for framework detection\n- Step 3 uses **curl via bash_tool** for outgoing link detection (method varies by framework)\n- Step 5 uses **Ahrefs MCP** (`site-explorer-top-pages`) for keyword research\n\n---\n\n## WORKFLOW\n\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\nSTEP 1 — DISCOVER ALL PAGES\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\n\nWHY THIS STEP EXISTS\nYou need a complete list of all blog\u002Fcontent pages to check. Curl-based sitemap\ndiscovery is fast, free, and reflects the live site — not a stale crawl cache.\n\nSUBSTEP 1A — FIND THE SITEMAP\n\nRun in order until one returns valid XML:\n\n```bash\ncurl -sL --max-time 15 \"https:\u002F\u002Fwww.DOMAIN.com\u002Fsitemap.xml\" | head -50\ncurl -sL --max-time 15 \"https:\u002F\u002Fwww.DOMAIN.com\u002Fsitemap_index.xml\" | head -50\ncurl -sL --max-time 15 \"https:\u002F\u002Fwww.DOMAIN.com\u002Frobots.txt\" | grep -i sitemap\n```\n\nIf none work, crawl the blog index directly:\n```bash\ncurl -sL \"https:\u002F\u002Fwww.DOMAIN.com\u002Fblog\u002F\" | grep -oP 'href=\"[^\"]*\"' | grep '\u002Fblog\u002F'\n```\n\nSUBSTEP 1B — EXTRACT ALL CONTENT URLS\n\n```bash\ncurl -sL \"https:\u002F\u002Fwww.DOMAIN.com\u002Fsitemap.xml\" \\\n  | grep -oP '(?\u003C=\u003Cloc>)[^\u003C]+' \\\n  | grep '\u002Fblog\u002F' \\\n  | grep -v '\u002Fblog\u002F$' \\\n  | grep -v '\u002Fpage\u002F' \\\n  | sort -u > \u002Ftmp\u002Fblog_urls.txt\n\nwc -l \u002Ftmp\u002Fblog_urls.txt\n```\n\nCall this LIST_ALL. Record TOTAL_PAGES = length of LIST_ALL.\n\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\nSTEP 2 — DETECT SITE FRAMEWORK\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\n\nWHY THIS STEP EXISTS\nThe detection method for outgoing links depends entirely on how the site renders HTML.\nRunning the wrong method gives completely wrong results — this step prevents that.\n\nRun both checks:\n\n```bash\n# Check response headers\ncurl -sI \"https:\u002F\u002Fwww.DOMAIN.com\" | grep -i \"x-powered-by\\|x-nextjs\\|x-vercel\\|server\\|generator\"\n\n# Check HTML source for framework signals\ncurl -sL \"https:\u002F\u002Fwww.DOMAIN.com\" \\\n  | grep -o 'name=\"generator\"[^>]*\\|__NEXT_DATA__\\|_next\\|__nuxt\\|ng-version\\|gatsby'\n```\n\nDECISION TABLE:\n\n| Signal found                          | Framework        | Method       |\n|---------------------------------------|------------------|--------------|\n| x-nextjs-prerender in headers         | Next.js          | Step 3A      |\n| _next or __NEXT_DATA__ in HTML        | Next.js          | Step 3A      |\n| x-vercel-cache in headers + _next     | Next.js\u002FVercel   | Step 3A      |\n| gatsby in HTML source                 | Gatsby (SSG)     | Step 3B      |\n| name=\"generator\" content=\"WordPress\"  | WordPress        | Step 3B      |\n| x-powered-by: PHP                     | Traditional CMS  | Step 3B      |\n| Static HTML (\u003Ca> tags present)        | Static\u002FSSG       | Step 3B      |\n| Zero \u003Ca> tags in curl response        | Unknown SPA      | Step 3C      |\n| __nuxt in HTML                        | Nuxt.js          | Step 3C      |\n\nIMPORTANT: Always test the detection method on 3 sample pages before running the\nfull list. Verify that a page you know has links returns links, and a page you\nknow is sparse returns few\u002Fnone. If results look wrong, re-read this step.\n\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\nSTEP 3 — DETECT OUTGOING INTERNAL LINKS PER PAGE\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\n\nRun the method determined in Step 2.\n\n────────────────────────────────────────\nSTEP 3A — NEXT.JS SITES (RSC HEADER METHOD)\n────────────────────────────────────────\n\nHOW IT WORKS\nNext.js with React Server Components exposes a server-rendered payload endpoint.\nBy sending `RSC: 1` as a request header, the server returns the full rendered page\ndata as a serialised React component tree — including all links — without needing\na headless browser. This works on all Next.js sites using the App Router (Next.js 13+).\n\nThis is the correct method for any site running on Next.js \u002F Vercel, regardless of\nwhether it uses ISR, SSG, or SSR. The RSC payload always contains the rendered\ncontent with all internal links visible as plain text paths.\n\nHOW TO VALIDATE THE FETCH WORKED\nA valid RSC response:\n- Is always larger than 10KB (most blog posts return 50KB–250KB)\n- Starts with React markers like `$Sreact.fragment` or `J:` or `1:\"$`\n- If a page returns less than 10KB or has no `$` in the first 500 chars,\n  the fetch failed — flag that page for manual review, do NOT mark it as a dead-end\n\n```python\nimport subprocess, re, time\n\ndomain = \"DOMAIN.com\"        # without www, e.g. \"infrasity.com\"\nblog_prefix = \"\u002Fblog\u002F\"       # content prefix, e.g. \"\u002Fblog\u002F\"\n\nwith open('\u002Ftmp\u002Fblog_urls.txt') as f:\n    urls = [u.strip() for u in f if u.strip()]\n\n# Asset extensions to exclude — these are not outgoing content links\nSKIP_EXT = (\n    '.png','.webp','.jpg','.jpeg','.gif','.svg','.ico',\n    '.woff2','.woff','.ttf','.eot','.otf',\n    '.css','.js','.jsx','.ts','.tsx',\n    '.xml','.txt','.pdf','.zip','.mp4','.mp3','.webm'\n)\n\n# Path prefixes to exclude — assets, CDN, internal Next.js paths\nSKIP_PREFIX = (\n    '\u002FPostImages\u002F', '\u002Fimages\u002F', '\u002Ffonts\u002F', '\u002Ffavicon',\n    '\u002F_next\u002F', '\u002F_vercel\u002F', '\u002Fstatic\u002F',\n    '\u002Fapi\u002F', '\u002F\u002Fcdn.', '\u002F\u002Fassets.'\n)\n\ndef get_outgoing_links(url):\n    result = subprocess.run([\n        'curl', '-sL', '--max-time', '10',\n        '-H', 'RSC: 1',\n        '-H', 'Next-Router-State-Tree: %5B%22%22%2C%7B%22children%22%3A%5B%22__PAGE__%22%2C%7B%7D%5D%7D%2Cnull%2Cnull%2Ctrue%5D',\n        url\n    ], capture_output=True, text=True, timeout=15)\n\n    data = result.stdout\n\n    # Validate RSC fetch succeeded\n    if len(data) \u003C 10000 or not any(m in data[:500] for m in ['$S', 'J:', '1:\"']):\n        return None  # Fetch failed — flag for manual review\n\n    # Extract all paths pointing to the same domain\n    # Extract both absolute domain links and relative paths starting with \u002F\n    # (excluding protocol-relative links starting with \u002F\u002F)\n    raw = re.findall(\n        rf'(?:(?:https?:\u002F\u002F)?(?:www\\\\.)?{re.escape(domain)}|(?\u003C!\u002F))\u002F(?!\u002F)([^\\\\s\\\"\\\\\\'\\\\\\\\),\\\\]]+)',\n        data\n    )\n    raw = ['\u002F' + r for r in raw]\n\n    clean = set()\n    for l in raw:\n        # Normalise: strip trailing junk\n        l = l.split('\\\\')[0].split('\"')[0].split(\"'\")[0].rstrip('),.]\\\\\u002F')\n        if len(l) \u003C 2:\n            continue\n        # Exclude assets\n        if any(l.lower().endswith(e) for e in SKIP_EXT):\n            continue\n        if any(l.startswith(p) for p in SKIP_PREFIX):\n            continue\n        # Exclude self-link\n        slug = url.rstrip('\u002F').split('\u002F')[-1]\n        if l.rstrip('\u002F') == blog_prefix.rstrip('\u002F') + '\u002F' + slug:\n            continue\n        clean.add(l)\n\n    return clean\n\ndead_ends = []\nhas_outlinks = []\nfailed = []\n\nfor i, url in enumerate(urls):\n    slug = url.split('\u002F')[-1]\n    try:\n        links = get_outgoing_links(url)\n        if links is None:\n            failed.append(url)\n            print(f\"[{i+1:3d}] FAILED   {slug}\")\n        elif len(links) == 0:\n            dead_ends.append(url)\n            print(f\"[{i+1:3d}] DEAD-END {slug}\")\n        else:\n            has_outlinks.append(url)\n            print(f\"[{i+1:3d}] LINKS({len(links):2d}) {slug}\")\n    except Exception as e:\n        failed.append(url)\n        print(f\"[{i+1:3d}] ERROR    {slug}: {e}\")\n    time.sleep(0.15)\n\nwith open('\u002Ftmp\u002Fdead_end_urls.txt', 'w') as f:\n    f.write('\\n'.join(dead_ends))\n\nprint(f\"\\nTotal:         {len(urls)}\")\nprint(f\"Dead-ends:     {len(dead_ends)}\")\nprint(f\"Has outlinks:  {len(has_outlinks)}\")\nprint(f\"Failed:        {len(failed)} (verify manually)\")\nprint(f\"Dead-end rate: {round(len(dead_ends)\u002Flen(urls)*100)}%\")\n```\n\n────────────────────────────────────────\nSTEP 3B — STATIC \u002F WORDPRESS \u002F TRADITIONAL CMS (STANDARD CURL)\n────────────────────────────────────────\n\nHOW IT WORKS\nFor sites that serve fully-rendered HTML directly (static files, WordPress, Gatsby,\ntraditional CMS), standard curl fetches the complete page including all links.\nParse only the main content area to exclude nav\u002Ffooter\u002Fsidebar links.\n\n```python\nimport subprocess, re, time\nfrom urllib.parse import urljoin\n\ndomain = \"DOMAIN.com\"\nbase = \"https:\u002F\u002Fwww.\" + domain\n\nwith open('\u002Ftmp\u002Fblog_urls.txt') as f:\n    urls = [u.strip() for u in f if u.strip()]\n\nSKIP_EXT = (\n    '.png','.webp','.jpg','.jpeg','.gif','.svg','.ico',\n    '.woff2','.css','.js','.xml','.pdf','.zip'\n)\n\ndef get_outgoing_links(url):\n    result = subprocess.run([\n        'curl', '-sL', '--max-time', '10',\n        '-A', 'Mozilla\u002F5.0 (compatible; InternalLinkBot\u002F1.0)',\n        url\n    ], capture_output=True, text=True, timeout=15)\n\n    html = result.stdout\n\n    # Extract main content area only\n    content = ''\n    for tag in ['article', 'main', 'body']:\n        m = re.search(rf'\u003C{tag}[^>]*>(.*?)\u003C\u002F{tag}>', html, re.DOTALL | re.IGNORECASE)\n        if m:\n            content = m.group(1)\n            break\n    if not content:\n        content = html\n\n    hrefs = re.findall(r'\u003Ca\\\\s+[^>]*href\\\\s*=\\\\s*[\"\\']([^\"\\'\\\\ ]+)[\"\\']', content, re.IGNORECASE)\n\n    clean = set()\n    for href in hrefs:\n        href = href.strip()\n        if not href or href.startswith('#') or href.startswith('mailto:') or href.startswith('tel:'):\n            continue\n        abs_url = urljoin(base, href)\n        if domain not in abs_url:\n            continue\n        path = '\u002F' + abs_url.split(domain, 1)[-1].lstrip('\u002F')\n        if any(path.lower().endswith(e) for e in SKIP_EXT):\n            continue\n        if abs_url.rstrip('\u002F') == url.rstrip('\u002F'):\n            continue\n        clean.add(path)\n    return clean\n\ndead_ends = []\nhas_outlinks = []\n\nfor i, url in enumerate(urls):\n    slug = url.split('\u002F')[-1]\n    links = get_outgoing_links(url)\n    if links:\n        has_outlinks.append(url)\n        print(f\"[{i+1:3d}] LINKS({len(links):2d}) {slug}\")\n    else:\n        dead_ends.append(url)\n        print(f\"[{i+1:3d}] DEAD-END {slug}\")\n    time.sleep(0.2)\n\nwith open('\u002Ftmp\u002Fdead_end_urls.txt', 'w') as f:\n    f.write('\\n'.join(dead_ends))\n\nprint(f\"\\nTotal:         {len(urls)}\")\nprint(f\"Dead-ends:     {len(dead_ends)}\")\nprint(f\"Has outlinks:  {len(has_outlinks)}\")\nprint(f\"Dead-end rate: {round(len(dead_ends)\u002Flen(urls)*100)}%\")\n```\n\n────────────────────────────────────────\nSTEP 3C — UNKNOWN \u002F PURE SPA — FLAG TO USER\n────────────────────────────────────────\n\nIf framework detection is inconclusive (e.g. Nuxt.js, Angular, Vue SPA) or if\ncurl returns fewer than 10 `\u003Ca>` tags across 3 sample pages, stop and tell the user:\n\n\"This site appears to use client-side rendering that cannot be detected through\nstandard curl requests. To run this audit accurately, one of the following is needed:\n\n1. Set up an Ahrefs Site Audit project for this domain (uses a headless browser\n   that renders JS). Share the project ID once the first crawl completes.\n2. Export a crawl from Screaming Frog with JavaScript rendering enabled and\n   provide the CSV of internal links.\n3. Confirm the exact framework being used so the right detection header can be applied.\n\nDo not proceed to generate a report until the detection method is confirmed working\non at least 5 sample pages.\"\n\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\nSTEP 4 — VALIDATE RESULTS\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\n\nWHY THIS STEP EXISTS\nDetection bugs are silent. A misconfigured asset filter or a failed fetch can\nsilently produce a wrong dead-end list. This step catches that before suggestions\nare generated.\n\nDo ALL of the following before proceeding:\n\n1. Pick 3 random pages from LIST_DEAD_ENDS. Open each in a browser and manually\n   confirm the page genuinely has no outgoing links to the domain. If any DO have\n   links, re-check the detection script's asset filter and domain pattern.\n\n2. Pick 3 pages from LIST_HAS_OUTLINKS. Verify they have visible links in the browser\n   to other pages on the same domain.\n\n3. If DEAD_END_RATE > 80%, something is likely wrong — the fetch may be returning\n   only asset links or the domain pattern may not be matching. Debug before continuing.\n\n4. If DEAD_END_RATE is 0%, the filter may be too loose (counting assets or self-links).\n   Check what links are being detected on known-sparse pages.\n\n5. For Step 3A (Next.js): if the failed list is non-empty, manually check those\n   pages. Some may genuinely be dead-ends; others may have errored. Re-run failed\n   pages individually before finalising the list.\n\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\nSTEP 5 — KEYWORD RESEARCH\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\n\nWHY THIS STEP EXISTS\nTo generate relevant outgoing link suggestions you need to know what each dead-end\npage AND each potential target page is about. One Ahrefs call covers the full blog.\n\nTOOL: `Ahrefs:site-explorer-top-pages`\n\nPARAMETERS:\n- `target`: blog prefix (e.g. `www.DOMAIN.com\u002Fblog\u002F`)\n- `mode`: `prefix`\n- `date`: today's date YYYY-MM-DD\n- `select`: `url,top_keyword,sum_traffic`\n- `limit`: 500\n- `order_by`: `sum_traffic:desc`\n\nRESULT: MAP_KEYWORDS — a keyword and traffic map for all pages.\n\nFor any dead-end page not in MAP_KEYWORDS, derive keyword from URL slug:\nreplace hyphens with spaces. NEVER fabricate traffic numbers — use null.\n\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\nSTEP 6 — CLUSTER AND GENERATE SUGGESTIONS\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\n\nWHY THIS STEP EXISTS\nOutgoing link suggestions must be topically relevant to the dead-end page. Clustering\nfirst, then matching within and across clusters, produces natural and useful suggestions.\n\nHOW TO CLUSTER\nGroup all dead-end pages by primary topic using their keyword and title. Create 5–15\nclusters depending on the site's content breadth. Adapt cluster names to the actual\ncontent — do not use generic names like \"Misc\" or \"Other\" unless truly necessary.\n\nHOW TO GENERATE SUGGESTIONS\nFor each dead-end page, identify 3 pages from LIST_ALL (any page on the site, not\nonly other dead-ends) that:\n\n1. Are topically related — same or adjacent subject matter\n2. Would be a natural next step for a reader of the dead-end page\n3. Are ideally already well-linked pages — so the dead-end connects into the\n   broader internal link graph\n\nFor each of the 3 suggestions, provide:\n\nANCHOR TEXT\nThe top keyword of the TARGET page (the page being linked TO). Rules:\n- Use the `kw` field from MAP_KEYWORDS for the target page\n- If target has no keyword data, derive from its URL slug\n- Must read naturally as hyperlinked text in a sentence\n- No em dashes (—) in or around the anchor text\n\nWHERE TO PLACE\nA specific section-level description of where in the DEAD-END page to insert\nthe outgoing link. Rules:\n- Section-level specific, not just \"somewhere in the article\"\n- Contextually logical — where a reader would naturally want to go deeper\n- Examples:\n  \"In the intro when defining the core concept\"\n  \"After the comparison table when suggesting next steps\"\n  \"In the Tools and Resources section at the end\"\n  \"When first mentioning [adjacent topic] mid-article\"\n\nSUGGESTED CONTEXT COPY\nA ready-to-paste sentence (1–2 sentences max) to drop directly into the dead-end\npage at the placement location. Rules:\n- Must contain the anchor text as a real HTML hyperlink to the TARGET page URL:\n  \u003Ca href=\"[target_url]\">[target_kw]\u003C\u002Fa>\n- Must read naturally in the context described in WHERE TO PLACE\n- Must add genuine value — not just \"click here to learn more\"\n- Should feel written for the dead-end page's specific voice and section\n- 15–35 words total (excluding HTML tags)\n- No em dashes (—) anywhere in the copy. Replace any em dash with a comma\n  or restructure the sentence entirely\n- Examples:\n  \"If you are just getting started, a clear overview of \u003Ca href=\"...\">developer\n   marketing strategy\u003C\u002Fa> will give you the framework to make these tactics work.\"\n  \"Teams that combine this with a solid \u003Ca href=\"...\">b2b content marketing\n   strategy\u003C\u002Fa> see compounding returns across both organic traffic and pipeline.\"\n\nStore this as the `copy` field in each suggestion object in D[].\n\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\nSTEP 7 — GENERATE HTML REPORT\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\n\nBEFORE WRITING ANY CODE read:\n  → references\u002Freport-style-reference.html\n\nThe reference file is the single source of truth for all CSS variables, typography,\ncomponent styles, spacing values, and JavaScript patterns. Do not invent new styles.\nDo not substitute different values. Copy them exactly from the reference.\n\nREPORT HEADER VALUES\n\n```\nSite URL:        https:\u002F\u002Fwww.DOMAIN.com\u002Fblog\u002F\nTotal Blogs:     TOTAL_PAGES\nNo Outlinks:     DEAD_END_COUNT        ← red stat (fail)\nHave Outlinks:   PAGES_WITH_OUTLINKS   ← green stat (pass)\nDead-End Rate:   DEAD_END_RATE%        ← amber stat (warn)\nSuggestions:     DEAD_END_COUNT × 3   ← indigo stat (accent)\n```\n\nFix-tag (choose based on method used):\n- Next.js:     `⚑ RSC payload detection (Next.js) · YYYY-MM-DD`\n- Static\u002FWP:   `⚑ Curl HTML parse · YYYY-MM-DD`\n- Site Audit:  `⚑ Ahrefs Site Audit (project ID: XXXX) · YYYY-MM-DD`\n\nREPORT FOOTER:\n`DOMAIN.com · Dead-End Pages Audit · YYYY-MM-DD \u002F Source: [detection method]`\n\nSUGGESTION CARD LABELS (different from orphan skill)\n- \"Link To\"        — the target page (page to link out TO)\n- \"Anchor Text\"    — the target page's top keyword\n- \"Where to Place\" — section in the dead-end page\n- \"Context Copy\"   — ready-to-paste sentence with link embedded\n\nDATA STRUCTURE — D[] array\n\nOne object per dead-end page:\n\n```javascript\n{\n  url:     \"https:\u002F\u002Fwww.DOMAIN.com\u002Fblog\u002Fpage-slug\",  \u002F\u002F dead-end page URL\n  title:   \"Page Title\",                              \u002F\u002F display title\n  kw:      \"top keyword\",                             \u002F\u002F dead-end page's own keyword\n  tr:      123,                                       \u002F\u002F monthly traffic or null\n  cluster: \"Cluster Name\",                            \u002F\u002F must match a key in CC{}\n  s: [\n    {\n      from:    \"Target Page Title\",                   \u002F\u002F page to link OUT TO\n      fromUrl: \"https:\u002F\u002Fwww.DOMAIN.com\u002Fblog\u002Ftarget\",  \u002F\u002F target URL\n      anchor:  \"target page keyword\",                 \u002F\u002F target's top keyword\n      place:   \"Where in the dead-end page to add the link\",\n      copy:    \"Sentence with \u003Ca href=\\\"...\\\">target keyword\u003C\u002Fa> embedded.\"\n    },\n    \u002F\u002F × 3 per dead-end page\n  ]\n}\n```\n\nNOTE ON FIELD SEMANTICS\n`from` and `fromUrl` refer to the TARGET page (the page being linked TO), not the\nsource page. This is the opposite direction from the orphan skill. The JS card\nbuilder must label these as \"Link To\" not \"Source Page\":\n\n```javascript\n\u002F\u002F In the suggestion card builder:\n`\u003Cdiv class=\"sug-from-label\">Link To\u003C\u002Fdiv>\n \u003Cdiv class=\"sug-from-title\">${s.from}\u003C\u002Fdiv>\n \u003Ca class=\"sug-from-url\" href=\"${s.fromUrl}\" target=\"_blank\">...\u003C\u002Fa>`\n```\n\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\nSTRICT RULES\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\n\n1. READ references\u002Freport-style-reference.html BEFORE writing any HTML or CSS.\n   Mandatory — no exceptions.\n\n2. ALWAYS run Step 2 (framework detection) before any link detection. Never assume\n   the framework based on the domain name or a visual inspection of the site.\n\n3. For Next.js sites, ALWAYS use the RSC header method (Step 3A). Never use standard\n   curl on a Next.js site — it returns an empty JS shell and produces wrong results.\n\n4. ALWAYS validate RSC fetch success per page. A response under 10KB or missing\n   React markers means the fetch failed. Flag those pages, never silently mark them\n   as dead-ends.\n\n5. NEVER count asset links (images, fonts, scripts, stylesheets) as qualifying\n   outgoing links. A page that only links to \u002Flogo.png and \u002Fstyles.css is a dead-end.\n\n6. NEVER count self-links (the page linking to itself) as a qualifying outgoing link.\n\n7. ALWAYS validate results on sample pages (Step 4) before generating suggestions.\n   If the dead-end rate is above 80% or exactly 0%, something is wrong — debug first.\n\n8. ANCHOR TEXT = top keyword of the TARGET page, not the dead-end page.\n\n9. NEVER use em dashes (—) in SUGGESTED CONTEXT COPY. Replace with a comma or\n   restructure the sentence.\n\n10. ALWAYS produce exactly 3 suggestions per dead-end page. No more, no fewer.\n\n11. ALWAYS make the dead-end page URL and all 3 target page URLs clickable\n    `\u003Ca href>` links opening in `target=\"_blank\"`.\n\n12. NEVER deviate from the CSS values in report-style-reference.html.\n\n13. ALWAYS build the report data-driven using D[] and CC[]. Never write static HTML\n    for individual blog cards.\n\n14. ALWAYS clean LIST_DEAD_ENDS before generating suggestions. Remove non-200 pages,\n    pagination, tag\u002Fcategory pages, and any URL that is not a standalone post.\n\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\nEDGE CASES\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\n\nNEXT.JS APP ROUTER vs PAGES ROUTER\nRSC header works on App Router (Next.js 13+). For older Pages Router sites, the\nRSC payload may not be available. Test signal: if response is under 10KB consistently\nacross multiple pages, the site may be using Pages Router. Fall back to checking\nfor __NEXT_DATA__ in the standard HTML response — this contains the serialised\npage props and may include link data.\n\nSITE USES A CDN THAT STRIPS HEADERS\nSome CDN configurations strip custom request headers before they reach the origin.\nIf RSC responses consistently return under 10KB, test by adding a cache-busting\nquery param: `?_rsc=1` appended to the URL. If that also fails, the CDN is likely\nstripping the RSC header — fall back to the __NEXT_DATA__ approach.\n\nLARGE SITES (200+ PAGES)\nFor sites with more than 200 blog pages, run the detection in batches of 50 with\na 0.3s delay between requests to avoid rate limiting. Prioritise the top-traffic\npages from MAP_KEYWORDS first. If time is a constraint, audit the top 100 pages\nby traffic and note in the report that the audit covers the top 100.\n\nALL PAGES ARE DEAD-ENDS (RATE = 100%)\nThis means the detection failed entirely — either the domain pattern is wrong or\nthe asset filter is too aggressive. Run the debug check: print the raw links found\non 3 known-linked pages. If nothing appears, fix the regex. If only assets appear,\nwiden the SKIP_EXT\u002FSKIP_PREFIX rules.\n\nDEAD-END RATE IS 0%\nEvery page appears to have outgoing links. The filter may be too loose — possibly\ncounting nav or footer links. Add `is_content` filtering by restricting the link\nsearch to only the `\u003Carticle>` or `\u003Cmain>` tag content area.\n\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\nCOMPUTING FINAL METRICS\n━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\n\nTOTAL_PAGES          = len(LIST_ALL)\nDEAD_END_COUNT       = len(LIST_DEAD_ENDS)\nPAGES_WITH_OUTLINKS  = TOTAL_PAGES - DEAD_END_COUNT\nDEAD_END_RATE        = round(DEAD_END_COUNT \u002F TOTAL_PAGES * 100)\nTOTAL_SUGGESTIONS    = DEAD_END_COUNT × 3\n",{},"main","38b32c99e7ed0b8e215d638e92cfa452c1ec3079","skills\u002Fimported\u002Fproductivity\u002Fno-outlinks-audit","https:\u002F\u002Fgithub.com\u002Fjonahthan433\u002FCortexOS","https:\u002F\u002Fgithub.com\u002Fjonahthan433","AI Consultant","2026-10-24 05:13:39",{"source":7,"skillName":5,"commands":30,"prompt":40,"zipAvailable":41},[31,34,37],{"tool":32,"command":33},"npx","npx skills add jonahthan433\u002FCortexOS --skill no-outlinks-audit",{"tool":35,"command":36},"bunx","bunx skills add jonahthan433\u002FCortexOS --skill no-outlinks-audit",{"tool":38,"command":39},"pnpm","pnpm dlx skills add jonahthan433\u002FCortexOS --skill no-outlinks-audit","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fjonahthan433\u002FCortexOS\u002Fblob\u002Fmain\u002Fskills\u002Fimported\u002Fproductivity\u002Fno-outlinks-audit\u002FSKILL.md",true,[43],{"path":44,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":47,"sourceHtmlUrl":48},"skill.json","REFERENCE","参考文档",297,"https:\u002F\u002Fgithub.com\u002Fjonahthan433\u002FCortexOS\u002Fblob\u002Fmain\u002Fskills\u002Fimported\u002Fproductivity\u002Fno-outlinks-audit\u002Fskill.json",[50,56,62,68,74,80],{"skillNo":51,"name":52,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":53,"categories":54,"keywords":55,"publishedAt":18},"01M7EDQZDQX3Q8Y7GCQX6KN9V7","opportunity-brief","https:\u002F\u002Fgithub.com\u002Fjonahthan433\u002FCortexOS\u002Fblob\u002Fmain\u002Fskills\u002Fsales\u002Fopportunity-brief\u002FSKILL.md",[],[],{"skillNo":57,"name":58,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":59,"categories":60,"keywords":61,"publishedAt":18},"01M7EDQYA0V9Y99VKBREJJ3MRT","morning-intel","https:\u002F\u002Fgithub.com\u002Fjonahthan433\u002FCortexOS\u002Fblob\u002Fmain\u002Fskills\u002Fresearch\u002Fmorning-intel\u002FSKILL.md",[],[],{"skillNo":63,"name":64,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":18},"01M7EDQVX1JVY0B6PDHKKGM19G","decision-brief","https:\u002F\u002Fgithub.com\u002Fjonahthan433\u002FCortexOS\u002Fblob\u002Fmain\u002Fskills\u002Fresearch\u002Fdecision-brief\u002FSKILL.md",[],[],{"skillNo":69,"name":70,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":18},"01M7EDQV1GENZAC2SZSD8KZ3GM","weekly-review","https:\u002F\u002Fgithub.com\u002Fjonahthan433\u002FCortexOS\u002Fblob\u002Fmain\u002Fskills\u002Fproductivity\u002Fweekly-review\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":18},"01M7EDQT2KETJ1Y3CGVN9EKW2J","system-health","https:\u002F\u002Fgithub.com\u002Fjonahthan433\u002FCortexOS\u002Fblob\u002Fmain\u002Fskills\u002Fops\u002Fsystem-health\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":18},"01M7EDQQN2JK3XPWMS1M4T0F3K","backup-restore","https:\u002F\u002Fgithub.com\u002Fjonahthan433\u002FCortexOS\u002Fblob\u002Fmain\u002Fskills\u002Fops\u002Fbackup-restore\u002FSKILL.md",[],[]]