Skip to content

Extend the Tier-0 rival-technology lexicon (FAMILIES) #1

Description

@michalstrnadel

The Tier-0 content scan (core/src/scorekeeper/detect/tier0_content.py) flags mid-turn when an agent writes a rival technology against a pinned decision — e.g. board pins attr:persistence.primary_db=postgresql and the agent writes import pymongo. It works off a small, high-precision lexicon of mutually-exclusive technology families (FAMILIES).

Task: extend FAMILIES (and _ALIASES where a package name differs from the tech, like pymongo→mongodb) to cover more of a real stack.

Good candidates: message queues (kafka/rabbitmq/redis-streams/sqs), cloud object storage (s3/gcs/azure-blob/minio), search (elasticsearch/opensearch/meilisearch/typesense), container/runtime, ORMs beyond the current set, JS bundlers, etc.

Acceptance:

  • new family/families added, lowercase, high-precision (no false-positive-prone generic words)
  • aliases added where the import token differs from the canonical name
  • a case in core/tests/test_cli_hooks.py::test_tier0_content_scan (or a new test) covering one new family
  • uv run pytest -q and uv run ruff check src tests pass

See CONTRIBUTING.md → "Good places to start". Small, self-contained, great first PR.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions