/verify-refs — Reference Authenticity Check

Running Python helpers

Choose the interpreter before running the examples. For a globally installed copy, use the private runtime recorded by its installer. In a checkout or linked workspace, use AWT_PYTHON when set, otherwise the toolkit's .venv (follow the skill directory link back to the toolkit): Scripts/python.exe on Windows, bin/python on macOS/Linux. Without that environment, check that python (Windows) or python3 (macOS/Linux) actually runs and has the helper's dependencies. Replace the example's python3 with that executable. In PowerShell, prefix a quoted executable with &; keep commands on one line and quote file paths.

Purpose

Check reference records for missing required fields, duplicate keys, malformed DOI values, malformed arXiv identifiers, and invalid URLs. The default mode is offline and deterministic.

Trigger Words

This skill activates on: verify refs, verify references, reference check, /verify-refs.

Workflow

1. Identify the target Markdown or .bib file. If the user does not specify one, ask.
2. Run: python3 .claude/skills/verify-refs/scripts/verify-refs.py --bib "{path}" --json
3. Report issues by entry key and severity.
4. For explicit metadata verification, run: python3 .claude/skills/verify-refs/scripts/verify-refs.py --bib "{path}" --json --online
5. Use CrossRef for DOI metadata, Semantic Scholar as a secondary metadata source, and arXiv for preprint identifiers. Online checks must be explicit because they depend on network availability.
6. In tests or offline review, use --metadata-dir "{dir}" to read CrossRef JSON, Semantic Scholar JSON, and arXiv Atom fixtures instead of live network calls.

Constraints

1. Do not import project-specific reference rules from other repositories.
2. Do not auto-fix reference records without user approval.
3. Keep output plain Markdown with no emoji.
4. Treat CrossRef, Semantic Scholar, and arXiv as verification sources, not as citation style authorities.