Inbound Anchor Risk Analyzer

Inbound Anchor Risk Analyzer

AGENT ROLE: Autonomous anchor-profile agent. Aggregate the inbound anchor distribution, classify it, assess manipulation risk, and emit the JSON in references/output.schema.json.

OBJECTIVE

Report the inbound anchor-text distribution for a domain (and optionally per target page), classify anchors, and flag unnatural exact-match/commercial concentration or velocity spikes that indicate over-optimization or link-scheme risk — with a recommended natural target mix.

INPUTS

• target (REQUIRED): the domain (or a specific URL) to analyze.
• brand_terms (REQUIRED string[]): to classify branded anchors correctly.
• money_keywords (OPTIONAL string[]): commercial terms to detect exact/partial commercial anchors.
• exact_ceiling (OPTIONAL, default 0.2): exact-match commercial ratio above which the profile is risky.

AUTHENTICATION (Backlink API)

1. REQUIRE DATAFORSEO_LOGIN+DATAFORSEO_PASSWORD or BACKLINK_API_KEY.
   • IF absent THEN STOP error.code="AUTH_MISSING_BACKLINK_PROVIDER".
2. Endpoint (DataForSEO): POST https://api.dataforseo.com/v3/backlinks/anchors/live with {target, limit, order_by:["backlinks,desc"]} → anchors with referring-domain counts.

EXPECTED TOOL CALLS

• Run scripts/inbound_anchors.py --target example.com --brand "Acme,Acme Inc".
• Fetch the anchors report (anchor text + referring-domain count + first-seen distribution when available).

PROCEDURE (deterministic)

STEP 1 — FETCH anchors weighted by REFERRING DOMAINS (not raw backlinks — sitewide links skew raw counts). STEP 2 — CLASSIFY each anchor: branded, branded_plus_keyword, exact_match_commercial, partial_commercial, generic ("click here", "website", "here"), naked_url, empty_or_image, other. STEP 3 — DISTRIBUTION: ratios by referring-domain weight; compute exact_commercial_ratio, branded_ratio, generic_ratio. STEP 4 — RISK FLAGS:

• OVER_OPTIMIZED_COMMERCIAL IF exact_commercial_ratio > exact_ceiling.
• LOW_BRANDED IF branded_ratio < 0.3 (natural profiles are branded-heavy).
• VELOCITY_SPIKE IF a large share of exact-commercial anchors share a tight first-seen window (link-scheme signal), when first-seen data is present. STEP 5 — TARGET MIX: recommend a natural distribution (e.g., branded 40–60%, generic/URL 20–30%, partial ≤20%, exact ≤5–10%). EMIT distribution, flags, risk_level, and the recommended mix.

RATE LIMITS & ERROR HANDLING

• Provider 429/402 → backoff 2^attempt (max 5) then STOP RATE_LIMITED/PROVIDER_PAYMENT_REQUIRED.
• IF the anchors report is empty THEN status="no_backlinks", no error.

MISSING / INSUFFICIENT DATA

• Weight by referring domains; IF only backlink counts are available THEN note weighting="backlinks" and caveat that sitewide links may inflate ratios.
• Velocity analysis requires first-seen data; IF absent THEN skip VELOCITY_SPIKE and note velocity_available=false.
• Risk is advisory — mark advisory=true; do not claim a penalty is certain.

OUTPUT

One JSON object per references/output.schema.json. No prose.

FILES

• scripts/inbound_anchors.py — anchors fetch, classification, ratio + risk analysis.
• references/output.schema.json — output contract.