Help

For AI Agents

1. llms.txt (simplest)

Every page is indexed in a machine-readable format:

curl https://rustman.org/llms.txt

Returns: all pages with titles, URLs, descriptions. Feed it to any LLM as context. ~19KB, 120+ entries.

2. Solo Factory skill (recommended)

npx skills add fortunto2/solo-factory/knowledge

The /knowledge skill loads llms.txt, finds relevant pages, fetches and synthesizes answers. Works with Claude Code, Cursor, Copilot, Gemini CLI, Codex.

/knowledge "how does harness engineering work?"
/knowledge "what stacks do we support?"
/knowledge "agent memory approaches"

3. Direct page fetch

Any page is plain HTML, works with WebFetch/curl:

WebFetch https://rustman.org/wiki/harness-engineering-summary
WebFetch https://rustman.org/posts/agent-memory-architecture

4. MCP (Solograph)

For deep integration — semantic search, code graph, session history:

claude mcp add -s project solograph -- uvx solograph

15 MCP tools: kb_search, session_search, codegraph_query, web_search, and more. See Solograph.

For Humans

Search

Click the search icon in the top nav (or press /). Searches all content. On section pages, scoped to that section.

Browse

• Course — structured learning path, 7 modules
• Posts — pillar articles (methodology, frameworks)
• Wiki — compiled knowledge pages
• Projects — build-in-public case studies
• Skills — 30+ executable Claude Code skills
• Stacks — technology stack templates

RSS

https://rustman.org/rss.xml

Content Structure

Built on Karpathy's LLM Wiki pattern: raw sources (immutable) → wiki (LLM-maintained synthesis) → blog (published subset). Files are source of truth, indexes are derived.