[phernandez]

Summary

Fixes #2005 by removing note formatting rather than restoring it. Breaking change.

Why it broke. 1.0 writes a note in two steps. First it accepts the note's bytes into note_content, which holds the canonical content and its checksum. Then LocalNoteContentStorage writes those bytes to the file. The formatter used to run inside FileService.write_file, which is no longer on that path, so format_on_save stopped running for bm_write and bm_edit.

Why it isn't restored. Formatting the file after the write would make it differ from the accepted note:

• The watcher would record the formatted file as an outside edit and accept a second version.
• The checksum the agent got back from the write would be stale, so its next bm_edit with expected_checksum would conflict.

Formatting before accept avoids that, but it means running a formatter subprocess inside the accept transaction. That's the wrong trade for a feature that's off by default.

Removed

• The config settings format_on_save, formatter_command, formatters and formatter_timeout, and their BASIC_MEMORY_* environment variables. Config uses extra="ignore", so existing config.json files that still have these keys keep loading.
• The bm format command and its man page entry.
• file_utils.format_file and format_markdown_builtin, and the formatter calls in FileService.write_file, FileService.update_frontmatter_with_result and MarkdownProcessor.write_file.
• The mdformat, mdformat-gfm and mdformat-frontmatter dependencies.
  • mdit-py-plugins came in through mdformat-gfm, and the page-citation integration test still imports it, so it's now an explicit dev dependency.
• The tests for these features. Three config tests used format_on_save or formatters only as sample settings; they now use skip_initialization_sync.

A CHANGELOG entry under Unreleased > Breaking Changes describes the removal.

Left in place

MarkdownProcessor still accepts app_config=, although nothing reads it now. basic-memory-cloud passes it in four places, so dropping the parameter would break the next cloud re-pin. Removing it should be a separate change, coordinated with cloud.

Verification

• just fast-check: clean
• Edited test files plus the man page and page-citation tests: 344 passed
• bm format now reports "No such command 'format'."

Follow-up

docs.basicmemory.com documents format_on_save and bm format, and needs updating when this ships.

🤖 Generated with Claude Code

https://claude.ai/code/session_01CMYVL4TcbWdXhdvAzxfqjo

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-05T00:00:51.227475Z	dfa6d3d	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.