Curate topics, tags, and tag taxonomy for canonical literature in `kb/library/literature/`, including refreshing `tags.yaml`, heuristic retagging from title and abstract, manual tag assignment, alias normalization, and taxonomy lint or apply passes. Use when Codex needs to maintain a paper tagging system after ingestion, clean up noisy tags, or prepare a better-tagged library for downstream literature analysis.
Use this after ingest whenever llm-wiki tag quality, taxonomy consistency, or retrieval cues need maintenance.
ingest (not owner): does not import PDFs or URLs.query (limited): read canonical metadata to derive candidate tags/topics.lint (owner): detect alias drift, noisy tags, and taxonomy schema violations.index (owner for tag-oriented projections): refresh tags.yaml, taxonomy projections, and tag-driven literature index views after tag updates.log (owner): preserve lightweight tagging audit notes in metadata.literature-corpus-builder) and deep note writing (research-note-author).metadata.yaml files under kb/library/literature/.taxonomy-sync after batch import so current canonical tags are represented in .tag-taxonomy.yamlretag for heuristic refresh from title plus abstract, or assign for manual updates.taxonomy-upsert to register canonical tags, aliases, topic hints, and descriptions.taxonomy-lint and taxonomy-apply to normalize alias and formatting drift.tags.yaml, taxonomy files, and tag-driven index views) so retrieval remains consistent, without taking over canonical source creation or duplicate handling.topics, tags, lightweight tagging notes, and taxonomy artifacts.canonical_title, authors, canonical_url, external_ids) during tag curation.tag-taxonomy.yaml.kb/memory/domain-profile.yaml.tag-taxonomy.yaml, tags.yaml, and per-source metadata.yaml notes), not left as ephemeral chat suggestions.literature-corpus-builder instead of patching around it here.python3 .agents/skills/literature-tagger/scripts/tag_literature.py refresh-index
python3 .agents/skills/literature-tagger/scripts/tag_literature.py taxonomy-sync --all
python3 .agents/skills/literature-tagger/scripts/tag_literature.py retag --all
python3 .agents/skills/literature-tagger/scripts/tag_literature.py retag --source-id lit-arxiv-2501-09747v1 --mode replace
python3 .agents/skills/literature-tagger/scripts/tag_literature.py assign --source-id lit-arxiv-2501-09747v1 --tag long-horizon --tag recovery --topic robot-learning
python3 .agents/skills/literature-tagger/scripts/tag_literature.py taxonomy-upsert --tag long-horizon --alias longhorizon --topic robot-learning --description "Long-horizon execution or recovery."
python3 .agents/skills/literature-tagger/scripts/tag_literature.py taxonomy-lint --all --strict
python3 .agents/skills/literature-tagger/scripts/tag_literature.py taxonomy-apply --all
cat kb/programs/my-program/workflow/reporting-events.yaml
tags.yaml as the only source of truth; canonical per-paper metadata stays authoritative.literature-corpus-builder.metadata.yaml entries for topics, tags, and tagging notes.tag-taxonomy.yaml when introducing new concepts or aliases.index.yaml, graph.yaml, and tags.yaml after batch operations.skill-evolution-advisor.