Import chapters

Importing chapters is the first step toward building structured canon.

Before you import

Check that:

• files are Markdown or plain text;
• filenames are stable and ordered;
• each file contains one clear chapter or section;
• the project scope is narrow enough to review;
• you have a backup of the original manuscript.

Import a small batch first

Start with a few chapters. A smaller batch makes it easier to see whether Arc is extracting the right entities, relationships, and evidence.

If the first batch works well, expand the project gradually.

Watch the ingestion stages

During ingestion, Arc may report progress for stages such as source staging, chapter extraction, semantic analysis, graph generation, and review preparation.

Treat warnings as meaningful. A completed-with-warnings result can still produce useful material, but the warnings explain which parts should not be trusted blindly.

After import

Open the project and inspect:

1. project status and warnings;
2. extracted chapters;
3. entity list;
4. relationship graph;
5. evidence and review items.

Do not judge the import only by whether it “finished.” Judge it by whether the extracted canon is reviewable and traceable.

When to re-import

Re-import when:

• the wrong files were uploaded;
• chapter boundaries were incorrect;
• the manuscript changed substantially;
• a warning shows that key semantic artifacts were not produced;
• the project is too noisy to repair manually.

If the product exposes retry actions, use them only when they are backed by real project state. Avoid creating multiple overlapping projects unless you intentionally want separate test runs.