daoyou · Spatial Archaeology Guide

What does daoyou · Spatial Archaeology Guide do?

What Is This

Turns "space as geological section" into an executable workflow. A user sends a place (place name / photo / one-sentence description), and the AI outputs a "Spatial Archaeology Archive": what it is now → what it used to be → how it changed layer by layer → what is buried under the labels → which people, events, and objects are worth telling.

Methodology origins: Song Nianshen's "Xita/Neighborhood" layered spatial history × Chongqing's Yixi lecture "The First Step of Being Brave" (sedimentary landforms / caves) × field investigation in Chengdu's Kuanzhai Alley.

Core Principles (Three Axioms)

1. The Present = Geological Section (Song Nianshen): Any spatial landscape seen today is merely the topmost layer of a historical mound, evolved layer by layer from the past, and the process never stops.
2. No Small Place, Only Un-excavated Places (Chongqing): Any insignificant corner is a "cave" — drill through it and you enter another world. Xita: fewer than ten streets, stacking 400 years of East Asian history.
3. Salvaging Complexity That Narratives Cannot Encompass (Chongqing × Xita): Nations/cities love commemorating "legitimacy," not "modernity"; labels, tags, and internet buzzwords are all post-hoc stickers — demystify them, dig out the buried layers.

Trigger

• Explicit command: daoyou Kuanzhai Alley, daoyou <a building/street/block>
• Natural language: "Guide me through XX", "Do an archaeology on XX", "How was the historical evolution of XX piled up layer by layer"

Execution Flow

Step 1: Lock Target and Current Status

• Define boundaries: single building → street → block/area; different levels require different archaeological depth
• Current status: function / form / fame; currently attached labels (tags: internet-famous street / food street / No.1 XX street / check-in spot…)

Step 2: Local First, Then Online

• Priority to local data: book lists, reading notes, interview and podcast transcripts, work log archives in your own database — your materials may already contain this place (books / transcripts / excerpts)
• Only search online when local data is insufficient

Step 3: Eight-Layer Archaeology (Universal Layered Framework, Tailored to Target)

From bottom to top, for each layer find "what traces/evidence this layer left in this space":

1. Natural Geography Layer: original landform (rivers / lakes / slopes / outside city walls), origin of the name
2. Belief/Religion Layer: pagodas / temples / churches / cathedrals / shrines — the earliest "meaning-installing devices," often the source of place names
3. Power/Governance Layer: imperial projects, administrative divisions, garrisons, government offices — who set rules here
4. Transport/Infrastructure Layer: railways / highways / canals / docks / city gates — transport determined where people settled
5. Economy/Industry Layer: workshops / factories / markets / commercial capital — rise and fall of industries
6. Population/Immigration Layer: who came, who left, settlement and diaspora, conflict and integration
7. Politics/Ideology Layer: wars / colonialism / reforms / movements / neoliberalism — how big eras crushed small places
8. Consumption/Symbol Layer: labels / tourism / internet-fame / reconstruction and imitation — today's skin

Step 4: Search and Evidence Collection (Source Discipline)

• Use available retrieval tools for vertical domain search (local history / local gazetteers / academic papers / news / archives); if the Agent has a specialized search skill, load it first
• When necessary, use web_extract to read original texts; prioritize local gazetteers, Wikipedia, and academic papers over self-media
• Record sources for each key historical fact

Step 5: Find "Threads" (Hooks)

A space is worth telling because concrete people / events / objects sew the layers together. Exemplary threads:

• A Korean-Chinese elderly person who lived in Xita for 70 years: one person's migration history = the living history of a street
• The Huanggutun bombing site is only 700 meters from the Xita Laman Temple: the physical intersection of two capitalist paths
• The WEST collider: a scientific milestone + witness to industrial decline, "stuck in a historical crack" (not old enough for a museum, not new enough to provide value)
• How to find: Who has lived here all their life? What major event happened right here? Which "fake antique" is hiding a real story underneath?

Step 6: Label Demystification

• When was this current positioning/name attached? What was it called before? What narrative was buried?
• Example: Xita = "Little Seoul" (attached post-1990s, a label for the city's use) → underneath is a Korean diaspora community; Northeast China ≠ backward — "it just experienced the future earlier"

Step 7: World-Historical Coordinates

Place local history back into world history: what was happening in the world at the same moment (e.g., 1904 Russo-Japanese War = the first modern artillery heard in the Xita wilderness). Let the "small place" reveal the weight of "big history."

Step 8: Final Output (Template)

# <Space Name> · Spatial Archaeology Archive
> Archaeology Date: YYYY-MM-DD | Level: Building/Street/Block | Trigger: daoyou

## I. Current Status and Labels
(What it looks like now, what tags are attached)

## II. Layer Section (Bottom-Up)
(For each layer: period + key traces + one-sentence evidence)

## III. Chronological Timeline
(Arranged by period; sources marked after facts)

## IV. Label Demystification
(When was the label attached, what was it called before, what was buried)

## V. World-Historical Coordinates
(Contemporary world events as background)

## VI. Threads (People/Events/Objects Worth Telling)

## VII. Pending Verification List ⚠️
(Facts not sourced go here; never fabricate)

## VIII. Source Index
(Each historical fact corresponds to a source: URL/local filename)

Discipline and Pitfalls

1. No Fabrication: Unsourced years/events go in the "Pending Verification List"; do not bluff with "probably" or "should be." Marking "Pending" = not in materials; verify with the person involved or check archives before external use.
2. Traceable Sources: Every historical fact in the archive must be marked with a source (URL/filename) for easy rechecking.
3. Search Discipline: Use dedicated search tools first; prioritize local data over online; prioritize vertical domain search over general search.
4. Primary Sources First: Original texts of local gazetteers / ancient books > modern summaries > tourism self-media. When any site contradicts an ancient text, use the text as the standard and record the conflict in the Pending Verification List (e.g., a tourism page says a temple "flourished in the Yuan Dynasty" but a Ming Dynasty local gazetteer says "built in the Chenghua period" — adopt the gazetteer, mark as doubtful).
5. Separate Legend from Historical Fact: Folklore (old monk saving Buddha / immortal riding a deer / prototype of a literary classic) must be recorded but clearly marked as "Legend Layer"; stories the author themselves noted as "may be fictional" should not be treated as contemporary history — "fictional narrative preserving something real" is itself a qualified thread; absolute truth is not required.
6. Delivery Format: Send the full long archive directly in the conversation (do not give file paths or send PDFs); maps / old photos can be attached as supplementary materials.
7. Save to Disk: After delivery, archive the file to the local archive directory (e.g., Spatial Archaeology/<Space Name>.md) and record it in the daily work log.
8. Continuous Archaeology in Same Valley: When archaeologically examining multiple targets in the same area (e.g., Sleeping Buddha Temple + Longjiao Temple in Cherry Valley), the second archive can directly reuse the search results from the first (same valley water system / same set of characters / same batch of local gazetteer sources) and mutually note "Related Archive" at the end of each — avoid re-searching the same batch of historical materials.
9. Reuse: The archaeological archive is "material," not the "final draft" — it can serve as raw material for subsequent writing, podcast outlines, or program introductions.
10. Shallow Object History, Dig into the Land: Late-stage businesses (nightclubs / shopping malls / factories / sports fields) often have only a few decades of their own historical records; archaeological depth comes from the land underneath. Treat adjacent or overlapping historical landscapes (temple fairs, old markets, former factory areas, temple precincts) as the "early form" of the object and write them into the industry/population layers. Point out the continuity of "same land, same function (letting ordinary people spend little money for a night's fun), different playlist" — this yields more archival value than a simple business chronology. Search order: first search the square/district name where the object's address is located + "history / evolution," then go back to search the object itself; apply the eight-layer framework as usual, but focus on the land-side layers.

Reference Cases

• Sleeping Buddha Temple (Beijing): Royal temple with six name changes and three recommendations of the Sleeping Buddha; demonstrates handling of "legend layer vs. historical fact layer"
• Longjiao Temple / Cherry Valley (Beijing): Temple built by eunuchs, two "Retired Elders"; demonstrates "continuous archaeology in same valley, mutual archive notes"
• Oriental Scala (Shenyang): Demonstrates "shallow business history, deep land history" — this single nightclub stacks back to early Qing temple fair grounds and 1920s mixed entertainment zones
• Paradigm Sources: Song Nianshen's "Xita/Neighborhood" layered spatial history, Chongqing's Yixi lecture "The First Step of Being Brave" (both publicly available; can be searched independently)

Version and Maintenance

• When content is updated, increment the version in the frontmatter and add a line version · date — what changed · why to references/CHANGELOG.md.
• Only distribute SKILL.md, README, LICENSE, and change log from the repository; archaeology archives and local materials are not checked in.