Show and tell: Obsidian High Recall for local-first vault recall in Codex

[ToussaintKnight]

I built Obsidian High Recall, a Codex skill / CLI wrapper for one problem: large Obsidian vaults can make normal search and agent memory silently miss relevant notes.

Project page: https://toussaintknight.github.io/obsidian-high-recall-skill/
Repo: https://github.com/ToussaintKnight/obsidian-high-recall-skill

What it does:

• runs locally over an Obsidian vault,
• reuses Smart Connections vectors when available,
• falls back to local hybrid/full-text search,
• returns broad recall packs for Codex when recall matters more than precision.

Privacy model: raw notes, private queries, snippets, and vault labels stay local. Public benchmark artifacts are aggregate/anonymized.

Fixture test:

git clone https://github.com/ToussaintKnight/obsidian-high-recall-skill.git
cd obsidian-high-recall-skill
npm test

Real-vault query:

node skills/obsidian-high-recall/scripts/obsidian_high_recall.mjs query "your broad research query" --vault "/absolute/path/to/your-vault" --backend auto --limit 120

I am looking for Windows/macOS/Linux testers with different vault sizes and Smart Connections states. The most useful report is whether it found notes that normal search or agent context would have missed, plus OS, Node version, Obsidian version, and Smart Connections status.

Please do not post private note paths, snippets, vault names, or private queries publicly.