/knowledge

Answer questions using the solopreneur knowledge base. Loads the site index via llms.txt, finds relevant pages, fetches and synthesizes answers.

# /knowledge

Answer questions using the solopreneur knowledge base. Loads the site index via llms.txt, finds relevant pages, fetches and synthesizes answers.

**Source:** [rustman.org/llms.txt](https://rustman.org/llms.txt) — full index of all published content.

## How It Works

The knowledge base is a static site (rustman.org) with 60+ pages organized as:
- **Posts** — pillar articles (manifesto, STREAM framework, harness engineering, agent memory, context graphs, launch playbook)
- **Wiki** — compiled knowledge pages (concepts, summaries, catalogs)
- **Projects** — build-in-public case studies
- **Skills** — solo-factory executable skills documentation
- **Stacks** — technology stack templates

All content indexed in `llms.txt` — a machine-readable map with titles, URLs, and descriptions.

## Steps

1. **Load index** — fetch `https://rustman.org/llms.txt` via WebFetch (or read from `references/llms-snapshot.txt` if offline)

2. **Parse question** from `$ARGUMENTS`:
   - If empty: show available topics from the index
   - If topic: find matching pages by keyword in titles/descriptions

3. **Fetch relevant pages** — for each match (max 3), fetch the full page:
   ```
   WebFetch https://rustman.org/wiki/{slug}
   ```
   Or read locally if in solopreneur repo:
   ```
   Read wiki/{slug}.md
   ```

4. **Synthesize answer** — combine information from fetched pages:
   - Answer the question directly
   - Cite sources with links
   - Note related pages for deeper reading
   - If the answer contradicts or extends existing knowledge, note that

5. **Suggest follow-up** — based on wikilinks in the fetched pages, suggest related topics.

## Usage Examples

```
/knowledge "how does harness engineering work?"
→ Fetches harness-engineering post + wiki summary, synthesizes key concepts

/knowledge "what's our approach to agent memory?"
→ Fetches agent-memory-architecture post + mempalace page, compares approaches

/knowledge "what stacks do we support?"
→ Lists all 9 stack templates with tech details

/knowledge "how to validate a startup idea?"
→ Fetches launch-playbook + SEED scoring + validation hub
```

## Offline Mode

If WebFetch is unavailable, the skill falls back to:
1. Local wiki files: `Grep "topic" wiki/` + `Read wiki/{match}.md`
2. Snapshot in `references/llms-snapshot.txt`

## Key References

- `references/llms-snapshot.txt` — cached copy of llms.txt (update with: `curl -s https://rustman.org/llms.txt > references/llms-snapshot.txt`)

## Content Domains

| Domain | Key Pages | Depth |
|--------|----------|-------|
| **Philosophy** | Manifesto, STREAM framework, antifragile design | Deep |
| **Launch** | Launch playbook, SEED scoring, kill/iterate/scale | Deep |
| **Engineering** | SGR, dev principles, CLI-first, privacy architecture | Deep |
| **AI Agents** | Harness engineering, agent memory, context graphs, RAG patterns | Very deep |
| **Tools** | Agent toolkit landscape, benchmarks, LLM providers | Catalog |
| **Projects** | 12 open source projects with case studies | Medium |
| **Skills** | 30 solo-factory skills with full documentation | Reference |