search-intent-coverage（搜索意图覆盖率）

Search intent coverage (SERP → outline)

You help the user build an outline that matches what Google is ranking and adds differentiated value.

Shared context first

Before asking repeated discovery questions, check whether .agents/seo-context.md exists.

If it does:

• read it first
• reuse the saved market, audience, topic, and competitor context
• ask only for the query or draft-specific details that are still missing

Inputs to collect

• Primary query
• Market (language + country)
• Optional: the user’s current URL/draft

Tooling & credentials

• Auth mode: mcp
• Requires: SERP API MCP
• Fallback: Browser MCP or agent-browser for reading competitor pages, plus user-provided competitor URLs or outlines if browsing is unavailable
• If missing: stop, ask the user to install or configure a SERP API MCP, and continue only after they confirm it is ready

Follow the shared setup and missing-access rules in docs/credentials-and-tooling.md.

Tools (adaptive)

Required: SERP API MCP. If it’s missing, stop and ask the user to install one.

For reading competitor URLs:

• Use a Browser MCP if available.
• Else use agent-browser CLI (install if needed).

Agent-browser install (if missing):

npm i -g agent-browser

Exact agent-browser commands to open a competitor page and extract common sections/headings fast:

agent-browser open "<url>"
agent-browser wait --load networkidle
agent-browser get title
agent-browser snapshot -c --json
# In the JSON, find the node with role="main" to get its @ref, then:
agent-browser get text @eMAIN
agent-browser snapshot -c -d 3 --json
# Scan the compact tree for role="heading" nodes to capture the H2/H3 pattern.

Fallback: ask the user to provide 3–5 competitor URLs and/or paste outlines.

Workflow

1. Fetch the SERP for the query (top 10).
2. Identify dominant intent type:

- informational / commercial / transactional / navigational

1. Extract patterns:

- repeated H2 topics - repeated definitions/FAQs - common “proof” elements (examples, templates, calculators)

1. Build a MECE outline:

- above-the-fold “answer fast” - core explanation - decision help (when commercial) - pitfalls, FAQs, examples

1. Add 1–2 differentiated elements (proprietary angle):

- checklist, interactive component idea, original data, teardown, template

Output

• “What people want” (1 paragraph)
• MECE outline (H1 → H2 → H3)
• “Differentiators to win” (3 bullets)