Opinionated behavioral layer for the Lore harness.
Lore OS adds guards, nudges, and knowledge capture behaviors to your agentic workflow:
- Memory Guard — Protects managed memory paths from direct agent access
- Harness Guard — Gates writes to
~/.config/lore/behind operator approval - Capture Nudges — Adaptive reminders to record findings during long sessions
- Ambiguity Scanner — Flags vague terms in prompts before they reach the agent
- Signal System — Nonce-stamped prompt directives with graduated color salience
Installed via lore init during first-run setup. Or manually:
lore bundle install lore-osThis clones the repo to ~/.lore-os/ and validates the manifest.
| Profile | Behavior |
|---|---|
off |
No hooks |
minimal |
Structural guards only |
standard |
All behaviors at default thresholds |
discovery |
Aggressive thresholds for exploration |
Set in .lore/config.json: { "profile": "standard" }
Fork this repo, install under a different slug (lore bundle install <your-slug> --url <your-fork>), and point your hook config at it. Scripts are vanilla Node.js — no build step, no dependencies.
Apache 2.0