About HList

HList is a directory of AI tools maintained by a small team. Every listing goes through the same pipeline: collected from public sources, verified against the official site, and published in four languages.

How a tool gets listed

1. Discovery. We track GitHub Trending, the GitHub API, HuggingFace Spaces, and public AI tool directories. Candidates are de-duplicated against the existing directory.
2. Verification. A tool is only published when its official website resolves, a screenshot of the live site is captured, and at least one data field (pricing, category, or description) is confirmed against the source.
3. Review queue. Listings marked unverified are in our review queue — visible in the directory but excluded from search-engine sitemaps until a human or automated pass confirms them. They are promoted gradually (30 per day) rather than dumped in bulk.
4. Publication. Verified listings appear in 4 languages (English, 简体中文, 日本語, 한국어), in the sitemap, and in the AI-agent data exits (llms.txt, tools.json, MCP).

What the data means

• Pricing tier — free / freemium / paid, based on the official pricing page. Every verified tool carries a pricing verified date; bulk listings show unverified until confirmed.
• China accessibility — whether the tool's site and service are reachable from mainland China without a proxy: ok (direct), unstable (intermittent), blocked (proxy required), or unknown (not yet checked). This is the dataset no other directory publishes.
• GitHub stars — live-fetched at build time when a tool has a public repository; historical snapshots are kept daily and power the weekly growth report.
• Community votes & comments — public API; seeded content is flagged in the database and tapered as real engagement arrives.

Editorial process

Descriptions are generated from the source's own metadata (official site, repository README, or HuggingFace card) and normalized for length and tone — we do not copy marketing copy from other directories. The pipeline runs daily at 06:00 UTC on a dedicated machine; every build is regression-checked before deployment.

AI assistance is used for content normalization and review-queue processing. The verification gates (website resolves, screenshot captured, data fields confirmed) are mechanical and applied to every listing, including AI-assisted ones.

Data exits for AI agents

• /llms.txt — index of machine-readable data
• /api/v1/tools.json — full structured dataset
• /api/mcp — MCP server (search, get tool, list categories)
• /datasets/ — downloadable datasets

Contact

Corrections and takedown requests: use the submit page or email [email protected]. DMCA: we comply with valid notices sent to the same address.