Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when planning link acquisition. Classifies the site's authority phase from site age and visible signals, then recommends phase-appropriate tactics from the bundled tactic playbook library. No backlink tool required.
.claude/skills/inhouseseo-linkbuilding/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 62% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 66% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 57% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 78% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 20% | 0% |
A phase-appropriate link acquisition strategy with specific, executable tactics. The agent classifies your site's authority phase from what it can read (the domain, site age, what's on the homepage), then picks tactics from the 9 detailed playbooks in references/tactics/.
Your domain or a URL (required). Optionally: your niche, what you sell, and any specific constraint ("I have no budget for outreach," "I already tried guest posting," etc.).
If the user specifies an action, adapt accordingly:
You are a senior link building strategist with 10+ years building backlink profiles across new, growth, and authority-stage sites.
Fetch the homepage of the domain. Google the brand name. From what you can see:
site:domain.com)Classify the phase:
State your reasoning. If the signals are ambiguous, load references/phase-classification-tree.md for the full decision tree before falling back on asking the user.
Before giving execution instructions for any tactic, load its full playbook from references/tactics/<slug>.md. The 9 tactics bundled in this repo:
entity-stacking.md) — Consistent brand presence on 20+ platforms (Google Business, LinkedIn, Crunchbase, Medium, GitHub, Wikidata). Expected: 20-30 referring domains month 1. The Wikidata entry is highest-leverage for entity recognition.citations-directories.md) — Industry and local directories. Chamber of Commerce links are DR 50-70, dofollow.competitor-backlink-gap.md) — Requires a backlink tool (Ahrefs/Moz). Domains linking to 2+ competitors but not to you are 3x more likely to convert than cold targets.guest-posting.md) — Target DR 30+ sites with real traffic. Contextual in-article links are worth 5-10x more than bio links.resource-pages.md) — Search intitle:resources [niche]. University pages (DR 70-90+) are gold. Pitch as gap-filler.skyscraper-technique.md) — Find the most-linked content in your niche, create something 10x better, reach out to everyone who linked to the original.strategic-partnerships.md) — Joint research, co-authored guides, integration pages. Integration page links are permanent and high-authority.podcast-guesting.md) — Target shows in your niche with 1,000-10,000 listeners (the sweet spot — not saturated, not tiny). Host links in show notes are dofollow and high-authority.new-site-launch-strategy.md) — The full 0-to-100 foundation playbook for a brand-new domain. Combines entity stacking, citations, first content, and realistic velocity.If the user can share anchor text distribution (from any backlink tool), compare against safe ranges:
If exact match > 5%: recommend branded/generic anchors for next 10-15 links before any outreach. For the full safety framework (why these ranges, what Penguin actually looks at, recovery playbook), load references/anchor-text-safety-guide.md.
If they can't share this data, skip this step.
Phase classification with reasoning.
Matched to the phase. For each: tactic name, why it fits this phase, expected referring domains, time investment, specific first action to take. Reference the full playbook.
Red flags: 50+ in one month, all from the same country, spike in exact-match anchors. Load references/link-velocity-redflags.md for the full red-flag catalog and SpamBrain detection patterns when diagnosing a velocity penalty.
Once foundation links are built, a keyword deep dive on your biggest opportunity keyword will show which pages would benefit most from the new link equity.
Load from references/ only when the step or situation calls for them.
Tactic playbooks (references/tactics/) — load one before giving execution instructions for that tactic in Step 2:
entity-stacking.md, citations-directories.md, competitor-backlink-gap.md, guest-posting.md, resource-pages.md, skyscraper-technique.md, strategic-partnerships.md, podcast-guesting.md, new-site-launch-strategy.mdDiagnostic and safety references (references/):
phase-classification-tree.md — decision tree when Step 1 signals are ambiguousanchor-text-safety-guide.md — full anchor safety framework and penalty recovery (Step 3)link-velocity-redflags.md — velocity red flags and SpamBrain patterns (diagnosing a traffic drop that looks link-related)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 20,040 | 17,975 | -10% | 1 | 1 | 0% | 3,063 | 4,437 | +45% | 0 | 0 | — |
case-02 | fail→fail | 23,406 | 19,096 | -18% | 1 | 1 | 0% | 3,340 | 4,502 | +35% | 0 | 0 | — |
case-03 | fail→fail | 21,454 | 18,294 | -15% | 1 | 1 | 0% | 3,595 | 4,633 | +29% | 0 | 0 | — |
case-04 | pass→pass | 12,706 | 10,412 | -18% | 1 | 1 | 0% | 2,073 | 3,427 | +65% | 0 | 0 | — |
case-05 | pass→pass | 7,154 | 9,897 | +38% | 1 | 1 | 0% | 1,323 | 3,545 | +168% | 0 | 0 | — |
case-06 | fail→pass | 13,886 | 12,141 | -13% | 1 | 1 | 0% | 2,231 | 3,623 | +62% | 0 | 0 | — |
case-07 | fail→pass | 13,923 | 10,434 | -25% | 1 | 1 | 0% | 1,984 | 3,287 | +66% | 0 | 0 | — |
case-08 | pass→pass | 20,515 | 14,509 | -29% | 1 | 1 | 0% | 2,833 | 3,898 | +38% | 0 | 0 | — |
case-09 | fail→pass | 20,126 | 20,305 | +1% | 1 | 1 | 0% | 3,005 | 4,708 | +57% | 0 | 0 | — |
case-10 | pass→pass | 17,782 | 16,661 | -6% | 1 | 1 | 0% | 2,553 | 4,149 | +63% | 0 | 0 | — |
case-11 | pass→pass | 14,544 | 18,109 | +25% | 1 | 1 | 0% | 2,267 | 4,331 | +91% | 0 | 0 | — |
case-12 | fail→pass | 16,797 | 18,830 | +12% | 1 | 1 | 0% | 2,472 | 4,388 | +78% | 0 | 0 | — |
case-13 | pass→pass | 9,675 | 8,196 | -15% | 1 | 1 | 0% | 1,602 | 2,991 | +87% | 0 | 0 | — |
case-14 | fail→fail | 15,396 | 15,054 | -2% | 1 | 1 | 0% | 2,378 | 3,998 | +68% | 0 | 0 | — |
case-15 | pass→fail | 23,138 | 17,742 | -23% | 1 | 1 | 0% | 3,434 | 4,322 | +26% | 0 | 0 | — |
case-16 | pass→pass | 15,328 | 10,312 | -33% | 1 | 1 | 0% | 2,415 | 3,208 | +33% | 0 | 0 | — |
case-17 | pass→pass | 18,185 | 14,459 | -20% | 1 | 1 | 0% | 2,608 | 3,901 | +50% | 0 | 0 | — |
case-18 | pass→pass | 15,486 | 15,320 | -1% | 1 | 1 | 0% | 2,098 | 3,825 | +82% | 0 | 0 | — |
case-19 | pass→pass | 14,007 | 9,769 | -30% | 1 | 1 | 0% | 2,030 | 2,939 | +45% | 0 | 0 | — |
case-20 | pass→pass | 19,650 | 15,055 | -23% | 1 | 1 | 0% | 2,866 | 3,787 | +32% | 0 | 0 | — |
case-21 | fail→pass | 14,232 | 6,657 | -53% | 1 | 1 | 0% | 2,158 | 2,590 | +20% | 0 | 0 | — |
case-22 | pass→pass | 8,908 | 8,842 | -1% | 1 | 1 | 0% | 1,440 | 3,111 | +116% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.