Skip to content

README.md

Latest commit

 

History

History
92 lines (68 loc) · 3.82 KB

README.md

File metadata and controls

92 lines (68 loc) · 3.82 KB
Raw

Data

The HackProof dataset: one YAML file per project, plus the config that produces it.

Layout

data/
  hackmeridian-2026.yaml         event config: window, awards, discovery queries
  integrations.yaml              Stellar integration partners and how each is detected
  schema-hackathon-project.yml   JSON Schema for every project file, generated
  schema-event.yml               JSON Schema for the event manifest, generated
  2026/_event.yml                event manifest: slug, name, network, window
  2026/<project>.yml             one file per HackMeridian 2026 project

A new event adds data/hackmeridian-2027.yaml. Point HACKPROOF_HACKATHON_FILE at it, run the pipeline, and export creates data/2027/. Nothing else changes.

Generating

From apps/api:

uv run hackproof export

This rewrites every project file for the event year, deletes files of projects that are gone, and rewrites the schema from the same models. Do not edit the project files by hand.

Publishing

The site publishes this folder at /data on every build:

  • /data/2026/<project>.yml
  • /data/schema-hackathon-project.yml
  • /data/hackmeridian-2026/projects.json and projects.yml: the whole year in one file, built from data/2026 so links shared earlier keep working. The project fields follow the project files, not the old export.

Event manifest

Each year folder holds _event.yml, which says what the files in it belong to: the event slug and name, the Stellar network it ran on (pubnet, testnet or futurenet), its starts_at and ends_at, and where the dataset is produced. A reader needs only this folder to import an event. The leading underscore keeps it from ever clashing with a project name.

File names

<project> is the repo name lowercased, with every run of characters outside [a-z0-9] replaced by one - (Stellar-Split becomes stellar-split). When projects in a year share a name, each gets its owner as a prefix, <owner>-<name>. A clash left after that stops the export.

Fields

Field Meaning
event Event slug, for example hackmeridian-2026
canonical_id pkg:github/<owner>/<repo>, the join key into other repo datasets
repo_url, name, owner Repo URL, repo name and owner login
description, homepage, language, license Repo metadata, left out when unknown
topics Tags the team set on the repo, left out when none
created_at Repo creation time, UTC
track genesis or scale
track_declared See below
award first or runner-up, left out when none
evidence_tier See below
commits_in_window Commits made during the event window
stack Stellar package to declared version, null when the manifest gives none
integrations Ecosystem partners found, each with evidence (method is package or endpoint, detail, file_url)
chain Testnet counts, the attributed accounts and the evidence behind the tier. Both lists are always present
contributors Public GitHub login and contributions. Always present

Stars, forks, total commits, last commit time, avatars and countries are not included. They change often or are personal.

evidence_tier

  • verified: the repo matches on-chain activity exactly (a deployer, contract or asset found in the repo), and that account was active on chain inside the event window.
  • likely: shown on the site as "Code only". The repo looks like a submission but has no on-chain proof.

track_declared

true only when the team states its track in its own docs, and track is then that value. false means the team did not say, and track is the default genesis, an assumption and not a fact.

Privacy

Public GitHub handles and public repo facts only. No emails, avatars, locations or other personal data.