SEO Content Engine

SEO Content Engine

A standing playbook that orchestrates the SEO sub-skills into a compounding engine. It is mostly agent orchestration + human checkpoints; the only bundled deterministic glue is a calendar-to-CSV renderer. Each phase reuses a sub-skill's scripts by path.

When to use

• "Build an SEO content strategy/engine for [client]", "what content should [company] be publishing?".
• A standing engagement, not a one-off report.

Sub-skills it orchestrates

• seo-content-audit — content inventory, performance signals, gap matrices, brand voice.
• aeo-visibility — answer-engine visibility for key queries (this is the env-gating sub-skill).
• topical-authority-mapper and/or programmatic-seo-planner — content architecture.
• A content-drafting step — done by you, the agent, primed on the brand-voice profile.

Invoke a sub-skill's scripts by relative path, e.g.:

python3 ${SKILL_DIR}/../seo-content-audit/scripts/crawl_sitemap.py --domain client.com ...
python3 ${SKILL_DIR}/../aeo-visibility/scripts/query_engines.py --queries-file prompts.json ...
python3 ${SKILL_DIR}/../topical-authority-mapper/scripts/expand_keywords.py --seeds "..." ...

How to run (phased, with checkpoints)

Phase 1 — Audit current state

Run seo-content-audit (inventory, performance, gap matrices, brand-voice profile) and aeo-visibility (answer-engine visibility for key queries). Output: a complete picture of where the client stands. [Human checkpoint after gap analysis — surface to the team channel, wait.]

Phase 2 — Identify content gaps

From the audit, surface competitive gaps, funnel gaps (missing TOFU/MOFU/BOFU), topic gaps, and comparison gaps. Prioritize by search volume × commercial intent × competitive difficulty (volume directional unless a paid API is connected).

Measured-metric enrichment propagates from the sub-skills. This playbook ships no enrichment adapter of its own — when DATAFORSEO_LOGIN/DATAFORSEO_PASSWORD are set, the seo-content-audit / topical-authority-mapper / programmatic-seo-planner sub-skills run their own bundled scripts/paid_seo.py (DataForSEO) in their optional enrichment step, upgrading volume/difficulty/authority/backlinks from "directional" to "measured". Those measured numbers flow up into this playbook's Phase-2 prioritization and Phase-4 calendar automatically. Absent creds → the keyless directional path is the default end-to-end; the adapter exits gracefully and nothing in the playbook breaks. (Semrush/Ahrefs are alternative providers, not implemented.)

Phase 3 — Build keyword architecture by funnel stage

Expand funnel-stage keyword variants and bucket them:

python3 ${SKILL_DIR}/../topical-authority-mapper/scripts/expand_keywords.py \
  --seeds "category, use cases, comparisons" --output ${WORKSPACE}/kw.json
python3 ${SKILL_DIR}/../seo-content-audit/scripts/serp_probe.py \
  --queries-file ${WORKSPACE}/kw.json --output ${WORKSPACE}/kw_serp.json

Bucket TOFU ("what is [category]", how-to) / MOFU (comparison, how-to-choose) / BOFU ("[Company] vs [Competitor]", "[Competitor] alternatives", pricing/migration); map each cluster to a content type.

Phase 4 — Create the content calendar

Sequence by urgency — BOFU-first (comparison/alternatives pages, especially if competitors run attack content), then MOFU, then TOFU + programmatic templates, plus 2–3 editorial/week ongoing. Build the calendar JSON, then render the deliverable:

python3 ${SKILL_DIR}/scripts/calendar_to_csv.py --input ${WORKSPACE}/calendar.json \
  --capacity 3 --period week --output ${WORKSPACE}/content-calendar-$(date +%F).csv

calendar_to_csv.py warns (stderr) on any week over --capacity. [Human checkpoint before drafting.]

Phase 5 — Draft content (you, the agent)

Per piece: match the client brand voice from the Phase 1 voice profile (never draft without it — it is the quality gate), weave target keywords naturally, build internal links to related content, add clear CTAs, include schema-markup recommendations. [Human checkpoint before publishing.]

Phase 6 — Internal-linking architecture

Design the link graph: TOFU→MOFU→BOFU→product/signup; all pages → relevant pillar.

Phase 7 — Publish & monitor (ongoing cadence)

Publish or hand drafts to the client. Track organic traffic by cluster, rankings by keyword (recurring serp_probe.py checks), and content→signup conversion. Wire a recurring trigger: weekly publish/monitor, monthly refresh, quarterly re-audit. Persist the metric series to Supabase if configured; deliver digests to the team channel.

Outputs

Phase artifacts to the workspace + team channel: audit report + brand-voice profile; gap analysis with priorities; keyword architecture by funnel stage; the prioritized content calendar (content-calendar-*.csv); drafted pieces in client voice with internal links, CTAs, schema recs; an internal-linking map; an ongoing performance dashboard.

Credentials / env

• Required (inherent): PERPLEXITY_API_KEY — querying answer engines is the task of the Phase 1 AEO sub-step (aeo-visibility), which hard-requires one answer-engine key; a sub-skill's hard requirement propagates to the playbook. (Any single answer-engine key the AEO sub-step uses satisfies this; Perplexity is canonical.) The additional engines are optional and degrade to whatever keys are present.
• Optional (each with a keyless fallback):
  • OPENAI_API_KEY/GEMINI_API_KEY/TAVILY_API_KEY — extra AEO engines; absent → that engine is skipped.
  • DATAFORSEO_LOGIN/DATAFORSEO_PASSWORD — if set → the sub-skills' bundled scripts/paid_seo.py returns measured volume/authority/backlinks; if not → SERP-derived directional signals (default; backlinks unavailable). SEMRUSH_API_KEY/AHREFS_API_TOKEN/ KEYWORD_ANALYSIS_API_KEY are alternative providers (not implemented in the adapter).
  • SUPABASE_URL/SUPABASE_KEY — performance-tracking history; absent → one-pass only.
  • channel/CMS creds (SLACK_* etc.) — auto-publish + digests; absent → drafts + workspace digest.
  • APIFY_API_TOKEN — hostile-site crawl fallback for the audit; absent → SERP + sitemap crawl.

Notes & edge cases

• Human checkpoints are mandatory after gap analysis (Phase 1), before drafting (Phase 4), and before publishing (Phase 5) — surface each to the team channel and wait for go-ahead.
• Brand-voice fidelity is the quality gate on drafts — always prime drafting with the audit's voice profile; never draft without it.
• BOFU-first sequencing — comparison/alternatives pages ship first; they convert fastest.
• Degrade gracefully: without a paid SEO/keyword API, prioritization leans on SERP-derived signals (Autocomplete demand, SERP rank, indexation) — flag volumes as directional.
• Ongoing cadence is the point — wire the recurring trigger for weekly publish/monitor, monthly refresh, quarterly re-audit; store metrics in Supabase to show compounding.
• Anti-block: route crawling/audit fetches through Robomotion Proxy + geo for volume.