Standalone CLI

The lore command runs the entire engine on your machine against a local database (no server, no account). It's bring-your-own-key — you supply an AI provider key, and Loretia never sends your story anywhere except to the model you choose.

Install it (macOS/Linux) with one line — it drops a single runtime-free binary on your PATH:

curl -fsSL https://loretia.com/install.sh | sh

This installs the command as lore, with loretia kept as an alias so older scripts keep working. (Re-running the installer is safe: it upgrades in place and cleans up any earlier loretia binary rather than leaving a duplicate behind.)

The CLI is self-describing. Run lore manifest for a machine-readable list of every command and flag (handy for scripts and for LLMs driving the tool).

Analyze a file, a folder, or a pattern

lore analyze chapter1.md                       # one file
lore analyze ./manuscript --sort date          # a whole folder, oldest first
lore analyze "scripts/**/*.fdx"                # a glob

Loretia figures out which files she understands (.fdx, .txt, .md, .pdf, .docx, .rtf, .odt, .epub, .html) and skips the rest. Use --sort name (default) or --sort date to order a batch.

Standalone vs. a shared project

By default each file is an isolated analysis — its canon never mixes with another's. To evaluate several files as one continuity (a novel's chapters, a franchise's scripts), group them into a project:

lore analyze ./novel --project "My Novel" --franchise "My World"

Bring your own key

lore analyze draft.txt \
  --provider openrouter \
  --model google/gemini-2.0-flash-lite-001 \
  --model-key sk-or-...

Providers: openrouter (default), anthropic, openai, google. Save defaults once with lore config --provider openrouter --model openrouter/free so you don't repeat them.

Organize your world

Loretia keeps a hierarchy of franchises → projects → scripts. Manage it directly:

lore franchise add "MCU"
lore project add 1 "Phase 1"        # project in franchise #1
lore franchise list --json
lore status --entities              # every script's status, score, and conflicts

Output formats

Pick how the report comes out with --format:

See Output formats for details and CI examples.