Shanhaijing mythic geography

Shanhaijing Mythic Geography

Treat the 山海经 as a layered compilation of mythic geography, creatures, ritual knowledge, and textual traditions—not as a literal modern world map. The skill can create a compelling interactive atlas, but every location must be labeled as textual, comparative, speculative, or fictionalized.

Runtime: Python 3.10+

The bundled validator uses the standard library only.

This is a learning and analysis skill. It should teach the user how a claim was acquired, what text supports it, what kind of inference it is, and what remains unknown. It must not turn a memorable myth, a modern illustration, or an attractive map into evidence.

Workflow

1. Identify the textual unit and edition: 山经, 海经, 大荒经, 海内经, or a later commentary. Record Chinese wording, translation, chapter, and source link.
2. Extract entities without flattening them: mountain, river, direction, distance expression, deity, creature, plant, mineral, ritual, and neighboring relation.
3. Preserve uncertainty. “East,” “inside,” “beyond,” and repeated distances are narrative/geographic conventions, not reliable modern coordinates.
4. Read textual-architecture.md and cross-text-boundaries.md before attaching a famous myth to the Shanhaijing.
5. Build maps in layers: textual topology first; comparative hypotheses second; artistic/worldbuilding layer last. Never draw a fictional alignment as an archaeological fact.
6. For a creature card, include name variants, original passage, attributes, habitat phrase, narrative function, later reception, and confidence label. Do not invent biological taxonomy.
7. Explain version differences and translation choices. End with open research questions rather than claiming to have found the real Kunlun or a literal global map.
8. When answering, use this order: evidence record → textual context → analysis → related-text boundary → uncertainty → next learning step.

Research assets

Read learning-path.md, entity-schema.md before creating a graph or database, creature-cards.md before writing a beast entry, and research-questions.md before proposing a modern map analogy.

The curated starter records live in data/. Check their source and claim layers before extending them. Run python3 scripts/validate_knowledge.py after changing any record.

Design rules

• Use arrows, rings, and regions for relationships that the text actually states; use dashed or clearly labeled lines for inference.
• Keep mythology, comparative geography, astronomy, and modern fantasy visually distinct.
• "World map turned vertical" can be offered as a creative analogy, never as proof that the text encodes a modern continent.
• Handle violent, monstrous, or discriminatory passages as historical material; do not reproduce them as claims about real peoples.
• 北冥与鲲鹏 are a Zhuangzi cross-text connection in this dataset, not a direct Shanhaijing creature claim.
• 共工触不周、天柱折 are recorded through the Huainanzi cross-text record; do not silently attribute the full story to the Shanhaijing.

Quality questions

• Can a reader return from every visible mark to a passage or an explicit inference?
• Are later popular images separated from early textual evidence?
• Does the map communicate uncertainty instead of false precision?
• Are names and translations preserved without copying long copyrighted editions?

Read sources.md before building a map or creature database.