For developers, researchers & AI agents

Build on Harpd data

Harpd publishes its product discovery and ranking data as open, machine-readable surfaces:datasets in JSON and CSV, a public read-only APIthat needs no key, and AI-readable files for answer engines. Everything derives from theHarpd Rank board and is licensedCC BY 4.0.

Quick answer

What can developers build on Harpd data?

Harpd exposes 1,122 products in the published Harpd catalog and every category board as JSON and CSV datasets, a read-only public API (no key required), and AI-readable surfaces (llms.txt, agent.json, the per-page SEO/GEO payload). All public data is licensed CC BY 4.0 — reuse commercially with attribution.

Products
1,122
Auth
None for public endpoints
License
CC BY 4.0
Schema
v1.0

Data source: Harpd public datasetsAPI referenceDataset indexMethodology

Why it matters

Open, key-less, CC BY 4.0 data lets any developer, researcher or AI agent build on Harpd Rank without negotiating access. Because every surface shares one schema and one licence, a tool built against one endpoint works against all of them.

Limitations

  • Public data reflects products submitted to Harpd Rank; it maps this platform’s coverage, not the entire software market.
  • The public API is read-only and rate-limited; it is for reuse, not for writing Rank data.
  • Rank Points are promotional placement, not an editorial quality score.

Data source: Harpd public datasets

Developer surfaces

Agent API

The purchasable machine surface: free search and rankings, paid comparison, research and model cost optimisation over x402. No signup, no API key, no dashboard.

Agent examples

Five worked examples — Claude, Cloudflare Agent, OpenAI-style client, generic HTTP agent and MCP agent — each walking discover → 402 → pay → retry → receive → cite.

Harpd for AI Agents

What an agent can buy, what it costs, and how payment works without a human in the loop.

Agent trust page

Methodology, freshness, provenance, versioning, uptime, privacy, limits and the refund policy an operator needs before connecting a wallet.

Public API

Every public endpoint: datasets, Rank API and the per-page SEO/GEO payload. No key, no signup, CC BY 4.0.

Open datasets

1,122 products in the published Harpd catalog and every category board as JSON and CSV, regenerated from live Rank data.

Dataset reference

Every dataset Harpd serves: endpoint, fields, update cadence and the exact licence terms for reuse.

Dataset catalogue

The full catalogue of public Harpd datasets with their scope, size and canonical URL.

Research reference

How Harpd research is produced and how to reproduce a report from the published snapshot and method.

Field schema

The field dictionary for every endpoint (schema version 1.0) and what the public dataset deliberately excludes.

Methodology

How Rank Points are created and ordered — the rules behind every number, machine-readable at /data/methodology.json.

Citations

Ready-to-copy citation lines (plain text and BibTeX) for every dataset, with the data date and coverage.

What is Harpd?

The canonical entity definition: what Harpd is, what it publishes and how AI systems access it.

Evidence register

Every public claim, its dataset, method and data date — the same register as /data/evidence.json.

AI / agent access

Machine-readable entity endpoints (products, categories, research, methodology, citations, Discovery Index) — JSON, no JavaScript.

Discovery Index

Non-sponsored coverage dataset of software, AI and developer products, independent of Rank Points.

External authority

Verified third-party references to Harpd — the one score Harpd cannot inflate itself.

Adopt Harpd data

Copy-paste datasets, API and embed snippets, the Powered by Harpd badge, and starter templates. Register your adoption in one click.

Adoption registry

Verified third parties building on Harpd data — the external-authority tracker for reuse.

Adoption metrics

Real adoption, citation and usage counts derived from Harpd data surfaces.

Machine-readable files

Data policy in one paragraph

The public datasets contain only public product and ranking information. They never include user accounts, email addresses, payment or Credit transaction data, moderation notes or unapproved submissions. Rank Points are promotional placement — paid with Credits — not an editorial quality score, and the data says so. The methodology is published at/rank/methodology/ and mirrored at/data/methodology.json. What is excluded is stated on the dataset index and the evidence register.

Citation-ready by design

Every dataset carries a ready-to-copy citation with its data date, coverage and licence — plain text and BibTeX on the citation hub. Quoting a single product? Link its profile page so the citation stays verifiable after the board moves.