Verwendung
- Installationsbefehl kopieren und das Quellpaket vor der Installation in der eigenen Agentenumgebung prüfen.
- Benötigte Werkzeuge, Zugangsdaten und Hosts im Original prüfen.
- Eingabe und gewünschtes Ergebnis beschreiben und die Ausgabe mit den Quellen abgleichen.
npx skills add https://github.com/venomous2/opencode-seo --skill seo-project-plannerVor dem Start
- Die Installation umfasst weder Modellabo noch API-Kontingent oder Dateizugriffsrechte.
- Sterne und Installationen zeigen Aktivität, keine unabhängige Qualitäts- oder Sicherheitsprüfung.
Aus dem Quell-Repository gerendert; Überschriften, Beispiele, Code, Tabellen, Links und Bilder bleiben erhalten.
SEO Project Planner
Turns a goal ("grow organic", "recover traffic", "launch a new site") into a phased engagement plan with clear deliverables — and optionally initializes the project's memory file so every later skill run has context.
Inputs
- Required: domain + primary goal
- Optional: timeline (default 6 months), budget/team constraints, known
competitors, site type (ecommerce, local, SaaS, publisher)
Setup
- Check for existing project memory:
python scripts/project_memory.py. - If none exists and the user wants persistent context, scaffold it:
python scripts/project_memory.py --init — then fill in domain, location, language, competitors, brand voice, and goals with the user's answers.
- Baseline reality check (keep it light — this is planning, not a full
audit):
python scripts/dfs_client.py ranked --target "<domain>" --limit 50
python scripts/dfs_client.py onpage --url "<homepage>"If credentials are missing, note it and plan anyway — but never fabricate baseline numbers.
Process
- Define success — convert the goal into measurable targets: organic
sessions, keywords in top 10, revenue/leads from organic, indexation coverage. Each target gets a number and a date.
- Phase the work — standard five-phase arc, adapted to the site type:
- Phase 1 — Foundation (weeks 1-4): crawl/index fixes, analytics and
Search Console verification, baseline benchmarks, keyword research.
- Phase 2 — Technical (weeks 3-8): architecture, canonicals, CWV,
schema, rendering. Depends on Phase 1 findings.
- Phase 3 — Content (weeks 6-16): priority pages, gap clusters,
refreshes. Depends on keyword research.
- Phase 4 — Authority (weeks 10-24): digital PR, link-worthy assets.
Starts only after content worth linking exists.
- Phase 5 — Iterate (ongoing): measure, refresh, expand.
- Milestones & deliverables — each phase ends with a named, checkable
deliverable: audit report, fixed-template deploy, content batch of N pages, asset + outreach campaign, quarterly report.
- Map dependencies explicitly — list what blocks what (e.g., "no link
outreach before 5 linkable assets live"). This prevents the classic mistake of promoting content that doesn't exist yet.
- Resourcing — per phase, note who's needed (dev, writer, designer,
SEO) and flag external dependencies (client approvals, dev release cycles).
Output
- Goal + measurable targets table
- Phase plan: phase | weeks | key activities | deliverable | depends on
- Milestone calendar
- Risk list (top 3, each with mitigation)
- Whether
seo-project.ymlwas created/updated - Single best next step (usually: run the baseline audit)
Long plans go to PROJECT-PLAN-<domain>-<date>.md.
Client deliverables: finish by running python scripts/report_publish.py <report>.md — it produces the branded HTML + PDF versions (plus executive one-pager variants) so every report is client-facing without extra steps.


