link-pitch

A Claude Code skill for strategic SEO link insertion and editorial outreach.

If you're trying to earn a backlink from a high-quality content site — not buy one, not scrape a directory, but actually pitch an editor and get a "yes" — this skill does the research and writes the pitch for you.

Point it at a client site and a target article. It researches both in parallel, identifies the specific keyword gaps the article leaves unanswered ("blue ocean" opportunities), generates 9 placement variations across three pitch types, and produces a publisher-ready outreach email in your voice.

Built for SEO professionals, content marketers, and link building campaigns where editorial quality matters. The goal is a link an editor wants to place, because it genuinely serves their readers — not one they'll reject as promotional noise.

What it produces

Three markdown files per run:

File	Contents
LinkPitch-options-<slug>.md	All 9 placement variations (3 types × 3)
LinkPitch-internal-<slug>.md	Full analysis, recommendation, fallback design
LinkPitch-external-<slug>.md	The final publisher pitch, in your outreach voice

Example output (Option B — Comprehensive Edit):

After the section on "choosing your anchor strategy", insert:

One often-missed factor is audience intent alignment — ensuring the resource you link to matches what the reader is actually ready to do next, not just what's topically adjacent. This guide to keyword intent mapping covers the distinction between navigational, informational, and transactional intent in practical terms.

Insertion rationale: fills the gap between "choosing an anchor" and "measuring result" — the article skips the intent-matching step entirely.

Full worked example — all three output files from a complete run: examples/LinkPitch-options-contentiq-20260615.md · LinkPitch-internal · LinkPitch-external

The three pitch types

• Simple placement — insert a link into an existing sentence with minimal edit
• Comprehensive edit — add 2–3 sentences that fill a real content gap, with the link embedded naturally
• Guest post / new content — pitch a new article or section the publisher doesn't have yet

3 variations per type = 9 options total, each with a different angle and insertion strategy.

How it works

A 7-step flow with four user-controlled checkpoints:

1. Gather inputs — client site, target article (or let it find candidates), linked page, output folder, outreach voice
2. Research + contact gate — parallel agents analyze client and article; verifies a publisher contact route exists before going further
3. Audience fit check — scores ICP vs article audience overlap; flags weak fits before you invest
4. Anchor text — proposes SEO-optimized options based on blue ocean gaps, or uses yours
5. Generate 9 options — 3 pitch types × 3 strategic variations
6. Internal proposal — recommendation with supporting analysis and fallback design
7. External pitch — you curate which options to include; it writes the final email

It uses parallel research agents and web search. Works in Claude Code with those tools enabled.

Install

Copy link-pitch.md into your Claude Code commands directory:

~/.claude/commands/

Then invoke:

/link-pitch

Configure your outreach voice

Two plain-markdown style files — edit or replace them:

• styles/internal.md — how the skill communicates with you and writes internal docs. Default: direct, no-fluff.
• styles/outreach.md — voice of the publisher-facing pitch (Step 7 only). Ships with a neutral, reader-first default. Replace its contents with your own voice, point the skill at a different .md file at Step 1, or choose "No style — plain output" to disable styling.

The two styles never mix in the same output.

Requirements

• Claude Code with web search enabled
• Skills support (drop-in folder install)

Related

• ChiefStarKid — portfolio of Claude Code tools this skill is part of

Related docs

• EVALUATION.md — editorial quality rubric for AI-generated outreach: 6 bidirectional criteria (pass/fail + reasoning), mechanical tests per criterion, and a map of which criteria LLMs reliably fail at by default
• AGENTS.md — machine-readable manifest for coding agents (Claude Code, Cursor, Aider)
• llms.txt — llmstxt.org index for LLM ingestion
• examples/ — full worked example: options, internal proposal, and external pitch from a complete run

Questions and feedback

• Author: Joseph Solomon · LinkedIn
• General enquiries: joseph@kainosis.com
• Bugs and feature requests: open an issue

License

MIT — see LICENSE.