SEO Fix Plan
Turn SEO audit findings into a concrete remediation plan: deduplicate issues into fix units, rank them by impact, confidence, effort, and dependency, and attach acceptance checks so the operator knows exactly what to change and how to verify it.
Agent-native
Run it in Claude, ChatGPT custom MCP apps, OpenClaw, Hermes, Codex, Claude Code, Cursor, VS Code, or another MCP-capable client. No dedicated GUI flow and no separate LLM API key.
Backed by live public data
Every step is grounded in live public-data records UnifAPI returns, so the output cites what is actually ranking, posting, or being said — not a generic best-practice list.
Composable & open source
Skills cross-reference each other and live in a public, MIT-style repo. Read the full SKILL.md on GitHub, fork it, or run it as-is inside your agent.
Paste this into Codex or Claude Code
The prompt is intentionally editable. Replace the handles, market, budget, and campaign goal, then let the agent call UnifAPI MCP when it needs live public data.
Turn this SEO audit for example.com into a prioritized fix plan. Group repeated issues into template-level tasks, rank quick wins vs blockers, and include acceptance checks for each fix.
The full skill, rendered from its SKILL.md
Turn SEO findings into an execution-ready fix plan. The job is not to run another generic audit; it is to take live SERP, rendered-page, backlink, and keyword evidence and sequence the smallest set of fixes most likely to improve crawlability, rankings, and click-through.
This is an enhanced skill: it reads live public data through UnifAPI when evidence is missing or stale, but it remains eyes, not hands. It does not edit the site, submit to Search Console, disavow links, or open PRs. It returns a plan the operator's own assistant or team can execute.
Use UnifAPI for live evidence
Start from an existing seo-audit report whenever possible. Re-pull only the evidence needed to confirm priority or fill a gap:
- Rendered page reality -
browser/markdown,browser/html, andbrowser/linksfor titles, meta, headings, JSON-LD, internal links, canonicals, robots directives, and orphan/depth signals. - Rank and competitor reality -
seo/serp,seo/competitors/domain,seo/competitors/relevant-pages, andseo/competitors/ranked-keywordsto see whether a fix targets a real ranking gap. - Keyword and intent fit -
seo/keywords/overviewandseo/keywords/intentto avoid optimizing a page for a query with weak demand or mismatched intent. - Authority and links -
seo/backlinks/summary,seo/backlinks/referring-domains,seo/backlinks/anchors,seo/backlinks/bulk-spam-score, andseo/backlinks/domain-intersectionwhen link risk or link gap changes priority.
Keep run dates and billing metadata in the output.
Workflow
- Load the diagnosis. Use the latest
seo-auditoutput if present. If the user only provides a domain and asks for fixes, run the minimum live reads needed to identify the current blockers. - Deduplicate issues into fix units. Merge repeated symptoms into one action when they share the same root cause, template, or component. A title-tag problem across 40 pages is one template fix, not 40 tasks.
- Classify every fix by type. Use: Crawl/indexation, Technical, On-page, Content quality, Internal links, Schema, Authority/links.
- Score priority. Rank by impact, confidence, effort, and dependency:
- Impact: blocked indexation or high-volume ranking gap beats cosmetic polish.
- Confidence: live evidence and competitor contrast beat best-practice hunches.
- Effort: template/sitewide fixes and striking-distance pages usually move first.
- Dependency: crawl/indexation fixes come before content and schema polish.
- Write exact acceptance checks. Every task needs a "done when" that can be verified with a live read, validator, or SERP re-check.
- Hand off cleanly. Point implementation details to
schema, the operator's code assistant, CMS, or site team. Do not pretend this skill made the change.
Priority rubric
Score each fix from 1-5:
| Factor | Meaning |
|---|---|
| Impact | Expected upside if fixed: indexation unlock, ranking lift, CTR gain, or reduced risk. |
| Confidence | Strength of evidence: rendered page, live SERP, competitor contrast, backlink data. |
| Effort | 5 = easy/template fix, 1 = heavy content or engineering project. |
| Dependency | 5 = unblocks other fixes, 1 = can wait. |
FixPriority = Impact * 0.4 + Confidence * 0.25 + Effort * 0.2 + Dependency * 0.15
Output tiers:
- Now - critical blockers and high-confidence quick wins.
- Next - material ranking or CTR fixes after blockers are cleared.
- Later - content expansion, authority campaigns, and lower-confidence bets.
Output
Return a concise execution plan:
# SEO Fix Plan - {domain} ({YYYY-MM-DD})
## Summary
- Highest-leverage fix: ...
- Main blocker: ...
- Quick wins: ...
- Evidence refreshed: ...
## Fix Plan
| Priority | Fix | Type | Evidence | Action | Acceptance check | Owner |
| -------- | --------------------------------------------- | ---------------- | ------------------------------------------------------ | -------------------- | -------------------------------------------------- | --------- |
| Now | Add canonical self-reference on product pages | Crawl/indexation | browser/html 2026-06-05: missing canonical on /pricing | Update page template | browser/html shows canonical; sitemap contains URL | Site/code |
## Implementation Notes
- Schema handoff: ...
- Content handoff: ...
- Link/outreach handoff: ...
## Re-check Plan
- Re-render affected templates after deploy.
- Re-run SERP checks for the target queries after the next crawl window.
- Record UnifAPI cost from billing metadata.
Guardrails
- Do not invent fixes that lack evidence. If evidence is missing, mark the task
needs-readand name the operation to run. - Do not promise rankings or timelines. A fix plan is a prioritized bet backed by live evidence.
- Do not mix implementation with diagnosis. This skill can produce exact copy, JSON-LD suggestions, or code-assistant instructions, but the operator's own tools make the change.
- Keep destructive SEO actions gated. Disavow, canonical consolidation, noindex, redirect changes, and URL removals need explicit human review.
Related Skills
- seo-audit: diagnose the site and produce the findings this skill turns into an execution plan.
- keyword-research: expand or validate the keyword gaps behind a content or on-page fix.
- schema: generate and validate JSON-LD for structured-data fixes.
- unifapi: the shared data skill (connect MCP, discover the SEO and browser operations this skill reads).
Source: seo-fix-plan/SKILL.md on GitHub — open a PR there to improve it.
The live APIs this skill calls
Every operation the skill names is one of these UnifAPI platforms — still visible and callable for product code, debugging, and custom agent flows.
- Prioritized fix table with type, evidence, action, owner, and acceptance check
- Quick wins separated from crawl/indexation blockers and longer content work
- Template-level grouping so repeated issues become one executable fix
- Re-check plan with the live operations to run after changes ship
More skills in the SEO Agent
Chain these in the same agent to go from one decision artifact to the next — each is its own run-prompt, workflow, and expected output.
SEO audit
Audit crawlability, on-page, and content against live SERP and ranking evidence — not a static checklist.
Open skillKeyword research
Turn a seed list into ranked keyword opportunities and competitor gaps from live SERP and volume data.
Open skillSchema
Choose and implement the right schema.org types for a page, with valid JSON-LD the assistant can write.
Open skillQuestions about SEO fixes
Pricing, workflow boundaries, public-data scope, and why this works better as an agent Skill.