# davidbmar.com — Project Portfolio

> AI-generated, machine-readable documentation for 132 of David Mar's GitHub
> projects. Each project covers what it is, how it's built (with architecture +
> sequence diagrams), and how to apply/reuse it, plus typed metadata
> (capabilities, tech, patterns, reuse_tags) for "have we built X?" queries.

## How to use this efficiently (progressive disclosure — do NOT load everything)

1. Scan the COMPACT index first (small — slug, title, one_liner, tech, reuse_tags):
   https://davidbmar.com/data/projects-index.json
   Use it to pick the few projects relevant to your question.
2. Fetch the FULL record for ONLY those slugs:
   https://davidbmar.com/projects/<slug>.record.json
   (full prose, Mermaid diagrams, components, depends_on, integrates_with, hygiene)
3. Human-readable page for a project:
   https://davidbmar.com/projects/<slug>.html
4. Browse all projects (humans): https://davidbmar.com/projects/

## Query interface for agents (search, don't download)

MCP server `ghps-mcp` exposes `portfolio_find_docs(query)` — returns ranked matches
(slug, title, one_liner, the fields that matched). Prefer this over downloading the
full corpus when you only need a few results.

## Full corpus (last resort — only if you genuinely need every record at once)

https://davidbmar.com/data/projects.json

## Field semantics

- reuse_tags, capabilities, patterns: best signals for "have we built X / can I reuse Y?"
- tech, depends_on, integrates_with: stack and dependencies
- thin=true: fork / stub / empty repo — low signal, may be skipped

Generated 2026-06-14T06:48:48.467336Z · 132 projects.