Skip to content

fix: make PubChem CAS resolution conservative - #7

Merged
rastagan-git merged 1 commit into
mainfrom
agent/conservative-pubchem-curation
Jul 19, 2026
Merged

fix: make PubChem CAS resolution conservative#7
rastagan-git merged 1 commit into
mainfrom
agent/conservative-pubchem-curation

Conversation

@rastagan-git

Copy link
Copy Markdown
Owner

Summary

  • Add repeatable --skip-pattern REGEX handling so structural workbook rows can be excluded before any PubChem request.
  • Separate provider execution from CAS curation with PubChem CAS Resolution, PubChem CAS Candidate Count, and configurable Resolved CAS output.
  • Populate a resolved CAS only for a confirmed CAS query or exactly one checksum-valid candidate; preserve multiple or missing candidates without choosing the first value.
  • Document the conservative workflow in both READMEs and the $curate-aroma-data skill.
  • Bump the package version to 0.3.1.

Root cause

The workbook workflow sent every identifier-shaped cell to PubChem and exported provider status together with raw CAS candidates. A valid PubChem record could therefore look like a resolved identity even when the row was structural, or when the record contained multiple or zero defensible CAS candidates.

Validation

  • ruff check .
  • ruff format --check .
  • pytest -q — 56 passed
  • Official skill validator — passed
  • Independent skill forward test — passed without network requests or file writes
  • Wheel build — aromanexus-0.3.1-py3-none-any.whl; canonical and legacy namespaces plus both console entry points verified

Closes #6

@rastagan-git
rastagan-git marked this pull request as ready for review July 19, 2026 14:57
@rastagan-git
rastagan-git merged commit ed90cde into main Jul 19, 2026
8 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Make workbook curation conservative for structural rows and multi-CAS PubChem results

1 participant