Open reasoning content behind Tiny Tars Foundation's health-literacy tools — the plain-language explanations that turn a person's own health data into something they can read, understand, and bring to a doctor's visit. It's structured so invited health-literacy experts can read and edit directly — this repo carries no app code, no credentials, and no PHI, so it can be shared with anyone doing that review.
- START-HERE.md — onboarding: editing reasoning (no code), or wiring the pipeline / adding a brain. Each track is ~5–10 minutes, hands-on.
- ARCHITECTURE.md — the frontmatter schema, lint rules, and sync hazards,
specified precisely. Read this when
validate-vault.shfails or you need the exact contract. - CONTRIBUTING.md — the structural-edit checklist and review process.
- CHANGELOG.md — version history: what changed and why.
A "brain" is a named dependency graph (a Dag, from the public @pablotech/neuro-pil engine):
what a piece of health reasoning is derived from, and what it feeds. finding-dag/ is the first
one — the graph behind {{PRODUCT_NAME}}'s Finding. A second brain later is a sibling folder at the repo
root, not a restructure of this one — see CONTRIBUTING.md.
Each brain folder is a plain-markdown Obsidian vault: one note per node. The exact frontmatter schema, section conventions, and lint rules are specified in ARCHITECTURE.md; the tutorial for opening one in Obsidian and reading its graph is in START-HERE.md.
*Generated by scripts/gen-graph-view.py (stdlib-only Python, no dependency to pin) directly from
finding-dag/*.md frontmatter, colored to match Obsidian's own dag/<kind> group palette — see
START-HERE.md § Color by kind if you want the
same coloring live in your own Obsidian vault. Re-run the script after a structural edit changes
the graph shape; it can never drift out of sync with the vault the way a hand-taken screenshot can.
Experts have full structural authority here — add, remove, or rewire nodes, not just edit prose on ones that already exist. Start at START-HERE.md to make a first edit, ARCHITECTURE.md for the exact schema and lint rules, and CONTRIBUTING.md before opening a PR.
Passing scripts/validate-vault.sh means the graph is well-formed — it is not a substitute for
expert review of the reasoning itself (see
ARCHITECTURE.md § What lint cannot check).
The private app that reads this graph syncs from its own checkout, not from here — see that repo's
own docs for how. validate-vault.sh and gen-graph-view.py (stdlib-only, no package.json here)
are the only scripts this repo ships or needs.
CC BY-SA 4.0 — see LICENSE. Copyright Tiny Tars Foundation. Reuse and adapt the
content freely, including commercially; credit the Foundation and release adaptations under the same
licence. It is a content licence, not a software licence, because that is what this repository holds.