mcpField
S
semantic-core-builder

semantic-core-builder

github.com/sergekostenchukMarketing & growth3250 installsUpdated 18w ago
Repository

About semantic-core-builder

A skill for turning keyword and search inputs into a structured semantic core, with build, merge, and audit modes. It creates machine-readable cluster data and a human-readable site structure, while recording evidence, confidence, and unresolved conflicts.

Key features

  1. 01Build, merge, and audit modes
  2. 02Intent- and topic-based clustering
  3. 03Cluster records with evidence and confidence
  4. 04JSON semantic core output
  5. 05Human-readable site structure

Use cases

  1. 01Structuring harvested keyword data
  2. 02Merging terms from multiple sources
  3. 03Auditing clusters for gaps or cannibalization

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/sergekostenchuk/ui-ux-agent-skill-system --skill semantic-core-builder

After installing, type/semantic-core-builderin Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

agents/
openai.yaml
assets/
semantic-core.schema.json
references/
semantic-core-contract.md

Details

From the project README

Semantic Core Builder

This skill turns raw keyword/search inputs into a structured semantic core. It hands the result to seo-llm-site-architect, concept-prototyper, and later design/implementation skills.

Operating Modes

  • build: create reports/semantic-core.json and docs/site-structure.md.
  • merge: merge existing-site keywords, user-provided terms, and harvested demand.
  • audit: review a semantic core for missing provenance, duplicate clusters, or weak intent mapping.

Required Workflow

  1. Read references/semantic-core-contract.md.
  2. Load reports/keyword-harvest.json or a user-provided equivalent.
  3. Cluster keywords by intent and topic; label any LLM grouping as Inference.
  4. Produce cluster records: intent, priority, canonical URL candidate, page type, evidence, and confidence.
  5. Write machine-readable reports/semantic-core.json and human-readable docs/site-structure.md.
  6. Hand off to concept-prototyper and seo-llm-site-architect.

Safety And Evidence

  • Do not invent search volume or SERP facts.
  • If no real SERP/source evidence exists, label the output as Manual or Inference.
  • Do not bury conflicts; record cannibalization, unclear intent, or missing evidence.

Handoff Shape

Input path:
Clusters:
High-priority URLs:
Conflicts:
Evidence gaps:
Output paths:
Next skills: concept-prototyper, seo-llm-site-architect

Validation

python3 $CODEX_HOME/skills/.system/skill-creator/scripts/quick_validate.py $CODEX_HOME/skills/semantic-core-builder
python3 $CODEX_HOME/skills/senior-skill-architect/scripts/lint_production_skill.py $CODEX_HOME/skills/semantic-core-builder

Comments

Markdown supported