Skip to content

copilot-instructions.md

Latest commit

 

History

History
43 lines (32 loc) · 2.92 KB

copilot-instructions.md

File metadata and controls

43 lines (32 loc) · 2.92 KB
Raw

Copilot / AI agent instructions for AstroML

This file contains concise, actionable guidance to help an AI coding agent be immediately productive in this repository.

Quick summary

  • AstroML is a Python research framework for dynamic graph ML on Stellar. Key workflows are ingestion, graph building, feature extraction, and training (see README).
  • Useful entry points: the command-line modules invoked in README such as astroml.ingestion.backfill, astroml.graph.build_snapshot, and astroml.training.train_gcn.

Discovery checklist (run first)

  • List top-level files: ls -la and tree -L 2.
  • Find build/test files: git grep -n "requirements.txt\|pyproject.toml\|setup.py\|Makefile\|Dockerfile" || true.
  • Find Python entrypoints and modules: git grep -n "^if __name__ == '__main__'\|def main\|module" || true.
  • Search for keywords used in README: git grep -n "backfill\|build_snapshot\|train_gcn\|config/database.yaml".

Project-specific patterns & conventions

  • CLI modules: Many workflows are exposed as python -m astroml.<submodule>. Prefer running those modules when reproducing workflows (examples are in README).
  • Config: Database and environment configuration live under config/ (README references config/database.yaml). Update/read these before running ingestion or tests.
  • Data flow: Ledger → Ingestion → Normalization → Graph Builder → Features → Models. When adding a change, map it to one of these stages.

Common commands (from README)

  • Create environment and install: python -m venv venv && source venv/bin/activate && pip install -r requirements.txt
  • Backfill ledger ingestion: python -m astroml.ingestion.backfill --start-ledger 1000000 --end-ledger 1100000
  • Build graph snapshot: python -m astroml.graph.build_snapshot --window 30d
  • Train baseline GCN: python -m astroml.training.train_gcn

Where to look for help in code reviews

  • When reviewing changes that affect ingestion, ensure database config and idempotency (re-run backfills safely).
  • For graph building changes, check windowing logic and reproducibility (random seeds, snapshot metadata).
  • For model/training changes, verify experiment reproducibility (config-driven, checkpoint paths) and alignment with README examples.

Integration & runtime hints

  • Expect PostgreSQL for ingestion; tests or local runs may require a local Postgres or a test fixture.
  • PyTorch / PyG are used for models — GPU availability changes performance but not required for basic runs.

When you can't find answers in the repo

  • If modules, tests, or config files are missing, run the discovery checklist above and ask the maintainer for missing artifacts (tests, requirements.txt, CI workflows).

Quick examples for the agent

  • To run the ingestion example locally: ensure config/database.yaml points to a local Postgres, then run the backfill command from the README.
  • To run a targeted search for tests: git grep -n "pytest\|unittest" || true.