During a shaping session,Documentation Index
Fetch the complete documentation index at: https://mintlify.com/xxyoudeadpunkxx/memory-assisted-shaping/llms.txt
Use this file to discover all available pages before exploring further.
notes.py generates three files. These are session-specific and should not be committed as part of the clean source package.
Default file locations
By default, all files are written next tonotes.py:
--dir to write session files to a different directory. See the command reference for details.
session_meta.json
Created byinit. Holds session identity for the duration of the shaping session.
If
session_meta.json already exists when init runs, it is validated — not overwritten. A missing or invalid session_id field raises an error and halts the command.session_notes.jsonl
The append-only live log. Each line is a compact JSON record. Records are never rewritten or removed.| Field | Description |
|---|---|
id | Zero-padded note ID (e.g. n00001) |
ts | UTC timestamp in ISO 8601 format |
session_id | UUID sourced from session_meta.json |
seq | 1-based sequence number |
type | One of the eight allowed signal types |
text | The signal text |
effect | (Optional) Only present when provided via --effect |
checksum | First 16 hex characters of the SHA-256 of the record, computed before the checksum field is added |
session_export.json
Produced by theexport command. Contains session metadata plus all records from the live log in a single JSON document.
The export file is a derived view.
session_notes.jsonl is the persistence source of truth. Do not treat session_export.json as canonical — re-run export to regenerate it if needed.Atomic writes
notes.py uses atomic writes for JSON files: it writes to a .tmp file, flushes, calls fsync, then renames the temporary file into place. The JSONL log uses append plus fsync per record. This protects against file corruption from interrupted writes — a partial write to the temporary file will not replace the prior valid file.
Command reference
Full reference for all five notes.py commands with flags, output shapes, and examples.
Back to Quickstart
Return to the quickstart guide.