SEOBot — OpenClaw SEO Research Agent

SEOBot is an autonomous OpenClaw SEO research agent that runs deep keyword analysis, live SERP breakdowns, competitor profiling, and content gap identification — then assembles everything into structured Markdown reports that give your content team a precise, data-driven edge.

It is not a content generator. It is a research analyst that never stops working: identifying the keywords worth fighting for, mapping the SERP you need to beat, exposing what your competitors rank for that you don't, and prioritizing every opportunity with a consistent scoring formula.

All research is powered by DataForSEO as the primary data layer, with optional integrations for Ahrefs (domain rating, backlinks), Google Search Console (your own site's actual rankings), and Serper (lightweight SERP scouting).

---

What It Does

Keyword Research (Mode 1) Expands any seed keyword into a full cluster using four DataForSEO Labs algorithms (suggestions, related keywords, keyword ideas, and category expansion), then enriches every candidate with search volume, CPC, competition, keyword difficulty, and search intent. A proprietary Opportunity Score (0–100) ranks every keyword across four weighted dimensions, tiering them into Priority / Strong / Moderate / Low action buckets.

SERP Analysis (Mode 2) Pulls a live top-10 SERP for any keyword using DataForSEO's Advanced endpoint. Detects every active SERP feature (featured snippet, PAA, video carousel, image pack, local pack, shopping, news box, sitelinks, knowledge panel), extracts People Also Ask questions, identifies the dominant content type across results, and maps which domains own which features. Ends with a concrete winning content strategy: format, length, angle, and PAA targets.

Competitor Research (Mode 3) Builds a full SEO profile of any competitor domain — their estimated organic keyword count and traffic, top pages by estimated visitors, and the 100 highest-value keywords they rank for that your site doesn't. Surfaces a prioritized attack plan ordered from most to least attackable based on competitor position, keyword volume, and difficulty.

Content Opportunity Report (Mode 4) Synthesizes all recent research into a prioritized content calendar: Quick Wins (publish in 2 weeks), Core Content (30 days), and Authority Builders (90 days). Cross-references with Google Search Console to find pages currently ranking positions 11–20 (one targeted optimization away from page 1) and pages with strong impressions but low CTR (title/meta optimization opportunities).

---

File Overview

seo-agent/
├── AGENTS.md             ← Research pipelines, scoring formula, quality rules, output formats
├── SOUL.md               ← Analytical philosophy and research character
├── SITE.md               ← ★ Your brief: domain, competitors, keyword clusters, preferences
├── HEARTBEAT.md          ← Weekly, bi-weekly, and monthly research schedules
├── REPORT_TEMPLATES.md   ← Full Markdown templates for all 4 report types
├── TOOLS.md              ← API reference with per-call cost estimates and rate limits
├── MEMORY.md             ← Research history, cluster coverage, rank tracking list
├── IDENTITY.md           ← Agent name and operator notification format
├── reports/              ← Generated Markdown reports (auto-created per research run)
├── research-queue/       ← Intermediate JSON data between pipeline steps
└── skills/seo-hub/scripts/
    ├── keyword-data.js         ← Volume + CPC + competition (DataForSEO Keywords Data API)
    ├── keyword-suggestions.js  ← Seed → up to 1,000 suggestions with KD + intent
    ├── related-keywords.js     ← "Searches related to" expansion (depth 1–4)
    ├── keyword-ideas.js        ← Category-based non-obvious discovery
    ├── keyword-difficulty.js   ← Bulk KD — 1,000 keywords per API call
    ├── search-intent.js        ← info / nav / commercial / transactional classification
    ├── serp-analysis.js        ← Live top-10 + 9 SERP feature types + PAA extraction
    ├── ranked-keywords.js      ← Competitor's full ranking keyword universe
    ├── domain-intersection.js  ← Keyword gap analysis
    ├── competitors-domain.js   ← Traffic estimates + top pages + competitor list
    ├── ahrefs.js               ← DR + backlinks + SERP overview (optional)
    ├── gsc-query.js            ← Your own GSC data — rankings, CTR, quick wins
    ├── serper-search.js        ← Fast SERP scouting fallback
    ├── score-keywords.js       ← Applies Opportunity Score formula, tiers results
    └── generate-report.js      ← Assembles and saves all report types as Markdown

---

Opportunity Score Formula

The scoring model is the core of SEOBot's research value. It ranks every keyword on a consistent 0–100 scale:

volume_score     = log10(volume + 1) / log10(1,000,000) × 40   [max 40 pts]
difficulty_score = (100 - KD) / 100 × 30                        [max 30 pts]
intent_score     = transactional:20 | commercial:15 | informational:8 | navigational:5
cpc_score        = min(cpc / 10, 1) × 10                        [max 10 pts]

Total (0–100):
  ≥70 = Priority   → attack immediately
  50–69 = Strong   → plan within 30 days
  30–49 = Moderate → 90-day plan
  <30 = Low        → monitor only

The logarithmic volume scaling means the jump from 1,000 → 10,000 monthly searches is not treated as ten times more valuable — it's approximately 1.3× more. This prevents high-volume head terms from drowning out high-intent long-tail keywords that are actually more winnable and more profitable.

---

Integrations

---

API Cost Profile

DataForSEO's pricing is per-call with no monthly minimum. A $10 credit covers months of typical usage for a single site.

---

Key Design Decisions

Four separate research modes, not one monolith. Keyword research is expensive if done wrong. By separating expansion → enrichment → scoring → SERP analysis into discrete pipeline stages, SEOBot only commits expensive SERP API calls (the most costly operation at $0.002/keyword) to candidates that have already passed the scoring filter.

Consistent scoring across all research. The same Opportunity Score formula is applied whether a keyword comes from seed expansion, a competitor gap, or a GSC quick-win scan. This lets SEOBot compare opportunities across completely different sources in a single prioritized content calendar.

Reports as the deliverable, not raw JSON. Every research run ends with a structured Markdown report saved to reports/. The agent never dumps raw API responses into chat. Strategists get executive summaries; the full data is in the tables.

SITE.md is the brief. All research is contextualized by the target domain, competitors, and keyword clusters defined there. This prevents SEOBot from producing generic research — every report references the specific site, niche, and content goals.

Cost transparency on every run. Every script prints its estimated API cost to stderr before running. SEOBot logs actual run costs to the daily memory file and respects MAX_MONTHLY_COST_USD — pausing autonomous runs and alerting the operator before spending overruns.