SuperAPI AI Directory

backlink-quality-validator

Marketing & growth

View repository

Updated 10-16 02:28

About this skill

Backlink Quality Validator

Use this skill after an opportunity is proposed by a scout or after a backlink already exists.

Read $HOME/SKILL/plans/seo-llm-skill-cluster/authority/white-hat-authority-policy.md before validating a placement.

Read references/validation-rubric.md before producing a verdict.

Owns

  • backlink quality review;
  • relevance and target-page fit;
  • link attribute review (follow, nofollow, ugc, sponsored);
  • anchor text risk;
  • spam/toxicity risk;
  • platform compliance notes;
  • approval or rejection recommendation.

Does Not Own

  • finding opportunities;
  • posting or editing links;
  • paid placement negotiation;
  • external outreach;
  • Search Console/rank attribution;
  • disavow decisions without broader SEO review.

Workflow

  1. Identify source URL, target URL, anchor text, link location, and context.
  2. Check relevance and user value.
  3. Check link attribute and disclosure expectations.
  4. Check platform-rule status.
  5. Identify spam, toxicity, PBN, directory, fake review, and paid-undisclosed risk.
  6. Score the backlink as approve, revise, reject, or needs_more_evidence.
  7. Provide exact reason and safe next step.

Non-Negotiables

  • Reject link farms, PBNs, fake reviews, hidden links, irrelevant directories, and paid-undisclosed links.
  • Reject anchor text that is misleading or stuffed.
  • Do not claim a backlink improves rankings without measurement.
  • Do not edit external pages.
  • Do not recommend disavow unless the broader context and risk are reviewed.

Safety And Privacy Boundaries

  • Do not store private outreach, account credentials, paid-placement negotiations, or private analytics in reports.
  • If source pages include private or sensitive content, summarize without exposing it.
  • Mark unknown platform rules as unknown rather than assuming permission.

Output Shape

Return:

  1. Validation verdict.
  2. Source/target/anchor/context.
  3. Scores and rationale.
  4. Link attribute and disclosure notes.
  5. Platform-rule status.
  6. Risks and required fixes.
  7. Monitoring or recheck date.

Validation

Validate skill edits with:

python3 $CODEX_HOME/skills/senior-skill-architect/scripts/lint_production_skill.py $HOME/SKILL/skills/backlink-quality-validator
python3 $HOME/SKILL/plans/seo-llm-skill-cluster/scripts/lint_skill_cluster.py $HOME/SKILL

Forward tests live in evals.json.