SEO Backlinks
by agricidaniel / claude-seo
Analyze a site's backlink profile, anchors, toxic signals, competitors, gaps, and disavow candidates.
Use it when you want a link building plan.
Listed ·Updated
Install data from skills.sh
Create Link Building Plan is a Link Building skill for AI agents, published in the seoskills.sh catalog. Reach for it when your work involves off-page signals, digital PR, and backlink outreach that holds up. Install it with one command and it runs inside your own agent, so the work happens in your workflow, not a separate SEO tool.
A link building strategy answers three questions before anyone sends a pitch: which pages need links, where competitors got theirs, and which tactics get the most links for the hours the team has. It hands back a 90-day plan, not a prospect list. Every tactic it picks ends the same way: a short list of sites, a person at each, and an address that works, by the shared outreach rules.
This skill runs on the manifold MCP server. Before the first step, confirm that its tools are available: look for seo_get_referring_domains and seo_get_backlink_summary (hosts often add a prefix, for example mcp__manifold__seo_get_referring_domains).
/mcp in Claude Code, or run codex mcp login manifold in Codex. Otherwise add the remote MCP server https://mcp.manifoldmcp.com/mcp and sign in: in Claude or ChatGPT as a custom connector, in Claude Code with claude mcp add --transport http manifold https://mcp.manifoldmcp.com/mcp and then /mcp. Other clients: https://www.manifoldmcp.com/docs/clients." Do not answer from memory instead.seo_* tools are there but the leads_* tools are not, the leads tool group is switched off on the app's Tools page (https://www.manifoldmcp.com/docs/authentication#turn-tool-groups-off). The plan needs no contacts; say so when a chosen tactic will.Before the first question, look for the product marketing context file: .agents/product-marketing.md (in older setups .claude/product-marketing.md or product-marketing-context.md). If it exists, read it and take every input it answers (the site, the competitors and their domains, the ICP, the pages that need links, the team and the budget) from it; ask only for what it lacks. If it does not exist and the job needs more than two answers about the business, offer manifold-get-started first.
Ask for these in one message. For anything the user does not answer, use the default and say that you assumed it.
seo_get_domain_overview in step 2 answers it if the user does not know.seo_get_serp_competitors that are real businesses, not publishers.seo_get_domain_overview on the site (5 credits) for domain rank, traffic and top pages; seo_get_backlink_summary on the site and on each competitor (10 credits each) for referring domains, dofollow share and broken backlinks; and seo_get_serp_competitors on the site (10 credits) only if the user named no competitors.seo_get_referring_domains with limit: 300 on each competitor and on the site (15 credits each). Count the domains that link to at least two competitors and not to the site, and how many of them clear the floors.seo_get_backlinks with limit: 300 on the strongest competitor (15 credits). Group the rows by url_to and name the three pages that earn most links and what they are (a tool, a study, a list, the homepage).seo_get_ranked_keywords on the site (10 credits). Keep the pages ranking 4 to 20 for keywords with volume and commercial or transactional intent; links move those first.broken_backlinks above zero in the baseline, or lost rows in the site's referring domains.seo_get_backlink_summary), new referring domains above the floor (seo_get_referring_domains with lost), links to the target pages (seo_get_backlinks on each page), and the position of the target pages for their keywords (seo_get_position, 6 credits each).Not using the CLI? Copy the SKILL.md and paste it straight into ChatGPT, Claude, or any agent.