# P2PWiki Graph Explorer — rebuild recipe

Revision: p2pwiki-20260909-83b16d

Reconstructed recipe; not the original prompts. End-to-end regeneration is untested.

Counts describe the captured snapshot, not current wiki coverage. Graph connections and summaries require provenance; this selection is not a full dataset backup.

## 1. Define the experience

Build a source-linked graph explorer for [KNOWLEDGE COLLECTION]. Inspect the repository and its instructions first, preserve existing work, and ask for the intended audience and available source material. Recreate the interaction idea rather than promising an exact copy of the original. Start with one complete, useful flow.

## 2. Prepare sources and assets

Create a versioned graph with stable node IDs, types, labels, source URLs, relationships and extraction provenance. Keep raw source references and distinguish editorial or model-derived connections from explicit source links.

## 3. Build the core flow

Implement search, suggested starting nodes, focus/fit controls, a neighbour list and a source-linked detail panel. Provide keyboard and text-list access independent of the graph canvas. Bound rendering and physics costs so large graphs do not exhaust memory.

## 4. Leave a working handoff

Use a restrained layout with readable typography and clear navigation. Check the production build, keyboard access, mobile layout, reduced motion, missing assets and network failures. Keep all key information available in HTML. Record what actually passed. Leave source, a lockfile, setup instructions, data provenance and unique assets in a small rebuild kit. Exclude secrets, dependency folders and caches. Capture a few representative moments for a showcase. Do not claim an exact backup or tested regeneration unless a restore was performed.