SEO Domain Analyzer

SEO Domain Analyzer

There is no Semrush/Ahrefs Robomotion package. The primary path is serp-derived ground truth (real, free SERP positions) via serp_probe.py, the SimilarWeb free page for a rough traffic estimate, and Wayback for traffic-trend signal. Paid SEO data is optional enrichment. You, the agent, label estimates vs measured and write the profile.

When to use

• "Pull SEO metrics for [domain]." / "How much organic traffic does [domain] get?"
• "Who are [domain]'s SEO competitors?" / "Where does [domain] rank for [keywords]?"
• The SEO data layer inside company-current-gtm-analysis or competitive landscape work.

How to run

1. Domain overview — indexed footprint (keyless)

python3 ${SKILL_DIR}/scripts/serp_probe.py site --domain example.com \
  --output ${WORKSPACE}/indexed.json          # sample of indexed pages + URL patterns

2. Keyword rank verification — the ground truth (keyless)

For each keyword, run a real SERP and record the domain's position, URL, and co-ranking competitors:

python3 ${SKILL_DIR}/scripts/serp_probe.py search --query "best rpa software" \
  --target-domain example.com --output ${WORKSPACE}/kw_rpa.json

target_rank, target_url, and co_ranking_domains are the trustworthy data points. If --keywords are not provided, infer them from the domain's content (homepage/blog) and your own knowledge, then probe each.

3. Traffic estimate — SimilarWeb free page (rendered)

npx playwright install chromium   # first run only
node ${SKILL_DIR}/scripts/render_page.mjs \
  --url "https://www.similarweb.com/website/example.com/" --wait 6000 \
  --output ${WORKSPACE}/similarweb.json

Read the visits/rank figures from paragraphs/headings. Label them estimates. If the page blocks, note reduced confidence and lean on the SERP signals.

4. Traffic trend (Wayback snapshot frequency)

python3 ${SKILL_DIR}/scripts/wayback_fetch.py --url https://example.com \
  --snapshots 5 --output ${WORKSPACE}/wb.json    # snapshot cadence ≈ activity/growth signal

5. Backlink signals (best-effort, keyless)

serp_probe.py search --query '"example.com" -site:example.com' surfaces who links/mentions the domain. Exact DR/referring-domain counts need a paid source (optional — see Credentials).

5b. (Optional) Enrichment — measured metrics (paid, off by default)

Everything above is keyless and yields a complete profile led by SERP ground truth. Only if DATAFORSEO_LOGIN/DATAFORSEO_PASSWORD are set, run the bundled DataForSEO adapter to replace the authority/traffic/backlink ESTIMATES with MEASURED numbers — this is the single biggest precision upgrade for this skill (the estimate columns become real):

python3 ${SKILL_DIR}/scripts/paid_seo.py domain --domain example.com \
  --output ${WORKSPACE}/paid_domain.json        # measured authority rank + organic traffic
python3 ${SKILL_DIR}/scripts/paid_seo.py ranked --domain example.com --limit 100 \
  --output ${WORKSPACE}/paid_ranked.json         # real ranked-keyword footprint + volumes
python3 ${SKILL_DIR}/scripts/paid_seo.py backlinks --target example.com \
  --output ${WORKSPACE}/paid_backlinks.json      # referring domains, backlinks, domain rank
python3 ${SKILL_DIR}/scripts/paid_seo.py keywords --keywords "kw1,kw2,kw3" \
  --output ${WORKSPACE}/paid_keywords.json       # measured volume/CPC/difficulty for targets

If creds are absent the adapter prints "paid enrichment unavailable — keyless path still applies" and exits non-zero without breaking this flow — skip it and keep the serp-derived profile. When the JSON IS produced, merge it into the profile and upgrade every "estimate"/"DR proxy" label to "measured" (authority/DR, organic traffic, keyword footprint, exact referring-domain/backlink counts). The SERP rank ground truth still leads. (Semrush/Ahrefs are alternative providers — not implemented in the adapter.)

6. Competitor discovery + comparison

Competitors = SERP co-rankers (step 2 co_ranking_domains) ∪ any provided list. Re-run steps 1–2 (lighter) per competitor and build a comparison table.

7. Assemble (you, the agent — no script)

Build the JSON profile + markdown summary: domain metrics (authority/DR proxy, est. organic traffic, keyword footprint, backlink signals, traffic trend), top pages, per-keyword rankings with SERP competitors, and the competitor comparison. Label every estimate as an estimate; lead with the SERP rank ground truth, not the guesses.

Outputs

• ${WORKSPACE}/indexed.json, kw_*.json, similarweb.json, wb.json — collected signal.
• ${WORKSPACE}/seo-profile-[domain].md (+ JSON) — the profile (your synthesis).

Credentials / env

• Required: none — the serp-derived path is keyless and produces a complete report.
• Optional: DATAFORSEO_LOGIN/DATAFORSEO_PASSWORD enable the bundled scripts/paid_seo.py enrichment (Step 5b) — measured authority/traffic/ranked-keywords/ backlinks that upgrade the estimate labels to "measured". APIFY_API_TOKEN (Semrush/Ahrefs public-page scraper actors) and a bring-your-own Semrush/Ahrefs key (no Robomotion package — alternatives, not wired into the adapter) are further options. All enrichment-only; the skill degrades to serp signals.

Notes & edge cases

• Traffic/authority numbers without a paid source are rough estimates — always label them; never present serp-derived guesses as exact metrics.
• SERP rank checks are the trustworthy ground truth; lead the report with those.
• serp_probe.py uses a keyless public HTML SERP endpoint that can rate-limit — it backs off; in a Robomotion deployment swap it for the robomotion-serp Search node (proxy + geo).
• Run competitors lighter (overview only) — full enrichment on many competitors is costly.
• SERP results are geo-personalized — note the country/language assumption in the report.