Build program-scoped literature evidence in a program's `evidence/` directory by reading the shared literature library. Use when Codex needs to summarize relevant sources, cluster themes, surface gaps and disagreements, or prepare grounded inputs for idea generation and idea review.
Use the canonical library as the evidence base for program-scoped llm-wiki synthesis.
ingest (not owner): does not import raw sources.query (owner): retrieve and rank canonical papers using charter cues plus tags/topics.lint (limited): validate evidence-map structure and citation trace completeness.index (limited): update program-local evidence artifacts, not global library indexes.log (owner): append report-ready analysis events for downstream weekly reporting.literature-scout) and canonical ingest (literature-corpus-builder).charter.yaml first.tags, topics, and short_summary.evidence/literature-map.yaml with , , , , and compact retrieval trace.ObservedInferredSuggestedOpenQuestionsworkflow/reporting-events.yaml.workflow/state.yaml to literature-analysis while preserving later-stage fields.research-landscape-analyst; if they need a human-facing reading bundle, route to research-deliverable-curator.inputs with charter plus literature entries/index artifacts that grounded the map.kb/memory/domain-profile.yaml as workspace-local query heuristic source.evidence/literature-map.yaml and reporting events), never kept only in chat.research-conductor can promote them into workflow/open-questions.yaml.python3 .agents/skills/literature-analyst/scripts/build_literature_map.py --program-id my-program
python3 .agents/skills/research-conductor/scripts/manage_workspace.py set-stage --program-id my-program --stage literature-analysis
cat kb/programs/my-program/workflow/reporting-events.yaml
research-note-author, research-landscape-analyst, or research-deliverable-curator.skill-evolution-advisor.