esg-taxonomy-tagging

name	esg-taxonomy-tagging
description	Classify ESG content across perspective facets (topic, industry, framework, jurisdiction, stakeholder). Use when the user provides ESG content to classify, categorize, or tag.
compatibility	opencode
metadata	{"role":"analyst","implementer":"opencode"}

Purpose

Classify ESG content across five perspective facets:

Facet	Description	Examples
topic	ESG subject area	climate-change, board-diversity, executive-compensation
industry	Sector applicability	financial-services, oil-gas, technology
framework	Reporting standard/framework	GRI, SASB, ISSB, TCFD, ESRS
jurisdiction	Geographic scope	EU, global, HK, US, CN
stakeholder	Primary audience	investors, regulators, corporates, consumers

Trigger Conditions

• User provides ESG content and asks to classify, tag, or categorize it
• User asks "what facets apply to this content?"
• User needs to identify frameworks, industries, or jurisdictions for a piece of ESG content

Required MCP Tools

• search_esg — search the KB for related pages to ground classification
• get_esg_metadata — retrieve valid sections, pillars, and source domains for taxonomy alignment
• list_esg_pages — browse pages by section/pillar to verify facet assignments
• get_esg_page — retrieve full page content for deep classification when needed

Input Schema

{
  "content": {
    "type": "string",
    "description": "The ESG content text to classify. Can be a full article, excerpt, or summary."
  },
  "context": {
    "type": "object",
    "description": "Optional hints to constrain classification.",
    "properties": {
      "known_topic": { "type": "string" },
      "known_industry": { "type": "string" },
      "known_framework": { "type": "string" },
      "known_jurisdiction": { "type": "string" },
      "known_stakeholder": { "type": "string" }
    }
  }
}

Output Schema

{
  "facets": {
    "type": "object",
    "description": "Classification results across the five perspective facets.",
    "properties": {
      "topic": {
        "type": "array",
        "items": { "type": "string" },
        "description": "One or more ESG topic labels."
      },
      "industry": {
        "type": "array",
        "items": { "type": "string" },
        "description": "Applicable industries/sectors."
      },
      "framework": {
        "type": "array",
        "items": { "type": "string" },
        "description": "Relevant reporting frameworks or standards."
      },
      "jurisdiction": {
        "type": "array",
        "items": { "type": "string" },
        "description": "Geographic or regulatory scope."
      },
      "stakeholder": {
        "type": "array",
        "items": { "type": "string" },
        "description": "Primary audience(s) for this content."
      }
    }
  },
  "confidence": {
    "type": "object",
    "description": "Per-facet confidence scores (0–1).",
    "properties": {
      "topic": { "type": "number" },
      "industry": { "type": "number" },
      "framework": { "type": "number" },
      "jurisdiction": { "type": "number" },
      "stakeholder": { "type": "number" }
    }
  },
  "supporting_pages": {
    "type": "array",
    "items": {
      "type": "object",
      "properties": {
        "page_id": { "type": "string" },
        "title": { "type": "string" },
        "section": { "type": "string" },
        "pillar": { "type": "string" }
      }
    },
    "description": "KB pages that support the classification."
  }
}

Procedure

1. If only a permalink is given, fetch the page via get_esg_page to obtain the full content.
2. Search the KB with search_esg using key terms extracted from the content to find related pages.
3. Call get_esg_metadata to obtain the taxonomy of valid sections, pillars, and source domains.
4. Map the content to the five facets, citing supporting KB pages for each facet assignment.
5. Assign confidence scores — lower scores when content spans multiple frameworks/industries or is ambiguous.

Worked Example: Classify "environmental/climate-change"

User request: "Classify the ESG Hub page on climate change."

Step 1 — Fetch the page

get_esg_page(page_id="environmental/climate-change")

Returns the full page content covering GHG emissions, Paris Agreement, TCFD, carbon pricing, transition risks.

Step 2 — Search for related pages

search_esg(query="climate change emissions TCFD carbon")

Returns related pages: environmental/carbon-accounting, frameworks/tcfd, environmental/ghg-protocol.

Step 3 — Get metadata

get_esg_metadata()

Returns sections list (environmental, social, governance, frameworks, ...) and pillars for alignment.

Step 4 — Classify

Facet	Assignment	Rationale
topic	["climate-change", "carbon-emissions", "transition-risk"]	Core subject is climate; covers emissions and transition
industry	["cross-sector"]	Climate applies to all industries
framework	["TCFD", "GHG-Protocol", "Paris-Agreement"]	Multiple frameworks referenced in content
jurisdiction	["global"]	Not jurisdiction-specific; global scope
stakeholder	["investors", "regulators", "corporates"]	TCFD targets investors; regulatory content targets all three

Step 5 — Output

{
  "facets": {
    "topic": ["climate-change", "carbon-emissions", "transition-risk"],
    "industry": ["cross-sector"],
    "framework": ["TCFD", "GHG-Protocol", "Paris-Agreement"],
    "jurisdiction": ["global"],
    "stakeholder": ["investors", "regulators", "corporates"]
  },
  "confidence": {
    "topic": 0.95,
    "industry": 0.70,
    "framework": 0.90,
    "jurisdiction": 0.85,
    "stakeholder": 0.80
  },
  "supporting_pages": [
    {
      "page_id": "environmental/climate-change",
      "title": "Climate Change",
      "section": "environmental",
      "pillar": "Environmental"
    },
    {
      "page_id": "environmental/carbon-accounting",
      "title": "Carbon Accounting",
      "section": "environmental",
      "pillar": "Environmental"
    },
    {
      "page_id": "frameworks/tcfd",
      "title": "TCFD",
      "section": "frameworks",
      "pillar": "Standards"
    }
  ]
}

Notes

• Facet values should align with the KB's existing section and pillar taxonomy from get_esg_metadata.
• If content does not fit cleanly into one value, return multiple and lower the confidence score.
• The supporting_pages array links the classification back to KB evidence for auditability.