Resources/Routines/Weekly SEO Brief

Weekly SEO Brief

weekly · scheduledclaude.ai routinesread-only3 tool calls

A Monday-morning brief that lands before your coffee does: the roadmap items that matter, how search moved last week, and the three things worth your time. A scheduled agent assembles it from your live Rampify data, so you start the week knowing where to look.

Build, don't copy

This page is a template, not a document. Don't fill in the blanks yourself. Copy the builder prompt below into your AI tool and it builds your personalized version with you: it resolves your project IDs over MCP, asks a few preference questions, and hands back Instructions ready to schedule.

What you get every week#

Under 400 words, four sections, every link deep into your dashboard:

Roadmap — highest priority outstanding
GSC week-over-week
What to work on this week
Top pages last 7 days

Build yours#

1
Launch it in your AI tool

The launch button opens your tool with the builder prompt already in the input. claude.ai, Cursor, and VS Code support prefilled deeplinks; for Claude Code it copies, ready to paste. The prompt is also copied to your clipboard every time, as a fallback. Your choice of tool is remembered for next time, and you can switch it from the dropdown.

2
Let it build with you

With the Rampify connector enabled (see Connecting), your AI calls list_projects to fill in your project, asks which day it should run, what to emphasize, and where the brief should land (email by default), then outputs your finished Instructions.

3
Schedule it

Paste the result into claude.ai → Routines → New routine and set the cadence you chose. The first run lands next Monday.

Tip

The conversation is the editor. Want a competitor section, or a 200-word version? Tell your AI while it's building, or come back any week and ask it to revise the Instructions.

The template#

Four parameters personalize it. The IDs come from list_projects, so you never hunt for UUIDs, and delivery is your call, made during the build:

ParameterTypeHow it's resolved
{{PROJECT_ID}}uuidrequiredThe project the brief reads. Returned by list_projects.
{{CLIENT_ID}}uuidrequiredScopes dashboard URLs (links use client_id, not project_id). Also from list_projects.
{{DOMAIN}}stringoptionalYour site. Lets you sanity-check the IDs landed on the right project.
{{DELIVERY}}instructionyour callWhere the finished brief goes. Your AI asks during the build: email via your email connector is the default, with Slack or just the run history as alternatives.
Instructions — template
You are a scheduled remote agent producing a Monday-morning SEO brief for the Rampify project. The audience is the founder, who will read this over coffee and decide what to focus on this week.

Project ID: {{PROJECT_ID}}
Domain: {{DOMAIN}}
Client ID: {{CLIENT_ID}}

The Rampify MCP server is attached. Use these tools (their actual names depend on the runtime; common forms include mcp__rampify__<name> or just <name>):

1. list_feature_specs — args: project_id={{PROJECT_ID}}, priority="critical,high", status="idea,planned,in_progress", limit=10, summary_only=true. Returns the highest-priority outstanding roadmap items.

2. get_search_performance — args: project_id={{PROJECT_ID}}, past 14 days. One call covers two sections: compute week-over-week deltas (most recent 7 days vs the prior 7), and take the top 5 pages by clicks for the most recent 7 days from the same response.

3. suggest_next_action — args: project_id={{PROJECT_ID}}, time_budget_minutes=60, limit=3. Top 3 opportunities to work on this week.

## Output

Produce a markdown brief, under 400 words total, with these sections in this order:

### Roadmap — highest priority outstanding
5–10 specs from list_feature_specs. Order: critical first, then high. Per row: title · status · dashboard URL (use the dashboard_url field returned by the tool — NEVER construct URLs from project_id manually). Flag any spec whose updated_at is >30 days ago as stale and worth triage.

### GSC week-over-week
2–3 sentences comparing the most recent 7 days vs the prior 7. Cover: notable click movers (up + down), notable average-position movers, and net WoW direction.

### What to work on this week
The top 3 from suggest_next_action. Per row: opportunity title · score · autonomy badge (agent / agent-draft-human-approve / human-only) · the suggested tool_to_use field. One-line rationale per item.

### Top pages last 7 days
List the top 5 pages with click counts (from the get_search_performance top-pages data). One-line note if any new page broke into the top 5 vs the prior week.

## Delivery

{{DELIVERY}}

## Constraints

- READ-ONLY against Rampify. Do not call any Rampify tool that modifies state (no update_*, create_*, link_commit, etc.). The Delivery step may use the email or messaging connector named above — nothing else.
- If a tool returns an error or empty result, note it in one line and continue with the rest.
- Stay under 400 words for the whole brief.
- The brief itself is the final output and is stored in the routine's run history — fully self-contained, even when it is also delivered elsewhere.
- Never construct dashboard URLs by hand. Always use the dashboard_url field returned by the tool. URLs use client_id, not project_id.
Read-only by design

This routine's constraints block forbids every mutating Rampify tool. Keep it intact when you customize: a scheduled agent should never be filing specs or editing keywords while you sleep. The one exception is delivery. The email or messaging connector named in {{DELIVERY}} may send the finished brief.

Tools this routine calls

Highest-priority outstanding specslist_feature_specsread
Search performance, week-over-week plus top pages, in one callget_search_performanceread
Top 3 opportunities for the weeksuggest_next_actionread

Plus delivery, if you chose any: your own email or messaging connector sends the finished brief per {{DELIVERY}}. That is not a Rampify tool.