Comparison table builder from raw data
Build a structured, embeddable comparison table from unstructured feature data.
Comparison tables are the highest-performing content format for tool-selection queries ('AGNT vs X', 'best AI agent platforms'). This prompt takes raw feature data (docs, marketing pages, changelogs) for 2-4 tools and produces a structured JSON table ready for the ComparisonTable component.
The prompt enforces factual accuracy by requiring a source URL for each cell. If a feature claim can't be verified from official docs, it's marked 'unverified' rather than stated as fact.
The prompt
<<<
You build a structured comparison table from raw feature data.
INPUTS:
- tools: list of 2-4 tool names with their docs URLs
- features: list of feature dimensions to compare
- (optional) raw notes or marketing copy to extract from
FOR EACH CELL:
1. Check the tool's official docs for the feature.
2. Rate: yes | no | partial | unknown.
3. Add a 1-sentence note explaining the rating.
4. Include source URL for verification.
OUTPUT:
{
"tools": ["AGNT", "Tool B", "Tool C"],
"features": [
{
"name": "A2A protocol support",
"ratings": {
"AGNT": { "status": "yes", "note": "...", "source": "..." },
"Tool B": { "status": "partial", "note": "...", "source": "..." },
"Tool C": { "status": "no", "note": "...", "source": "..." }
}
}
],
"last_verified": "2026-04-12"
}
RULES:
- Never claim a tool lacks a feature without checking docs first.
- "unknown" is better than a wrong "no".
- If a tool's docs are gated or unavailable, mark all its cells "unknown".
- Include last_verified per table, not per cell.
>>>When to use
Use this when building comparison or 'alternatives' pages. The structured JSON output plugs directly into the frontend ComparisonTable component.
Works with
Related guides
Share as social post
Comparison table builder from raw data (claude-code) — Build a structured, embeddable comparison table from unstructured feature data. https://agntdot.com/prompts/comparison-table-builder
188 / 280 chars
Related prompts
Venue intake prompt pack for Claude Code
Parse a venue's website + menu into AGNT's knowledge pack format.
Ecosystem release-teardown prompt
The prompt the AGNT signals desk uses to draft ecosystem release breakdowns.
MCP graph-query template
Query AGNT's knowledge graph through MCP with clean fallbacks.