Memos Overview

Memos (quick notes) are the fifth file entity in Backlog.md, designed for one-off casual notes: a meeting conclusion, a critical log snippet during debugging, or a fleeting idea. Previously anything worth remembering had to become a task or document—even a single-line meeting record had to fit the ID scheme, title, and paragraph structure. Memos make "write it down now, decide later" possible.

Differences from Other Entities

Tasks / Drafts Documents / Decisions Memo
Purpose Executable, trackable work items Formal knowledge assets Lightweight capture inbox
ID back-123 / draft-3 doc-5 / decision-3 20261003-1 (date + daily sequence)
Title Required Required No title; derived from first body line
Structural requirements Status, acceptance criteria, and other metadata Type, directory organization Body only + optional tags

Memos are deliberately lightweight: no status flow, no milestones, no acceptance criteria, and they do not enter the dependency graph (Graph View). Their value lies in outbound connections—bare entity IDs and path written in the body automatically become links, turning casual notes into growable nodes of the knowledge network (see Search & Knowledge Linking).

Storage Format

All memos are saved under the backlog/memos/ directory, one Markdown file per note:

---
id: 20261003-1
created_date: 2026-10-03 01:14
updated_date: 2026-10-03 01:14
tags: [meeting]
---

Weekly meeting conclusion: v2 release delayed by one week; fix performance regressions first. Related task back-123.
  • ID rule: YYYYMMDD-N format; N is the max existing memo sequence of the day + 1, resetting to 1 across days.
  • Frontmatter: only id, created_date, updated_date, and optional tags (omitted when the tag list is empty); there is no title field.
  • Card title: derived from the first non-empty line of the body; if the body starts with blank lines, falls back to the first 40 characters.
  • Temporary data: backlog/memos/ is treated as a temporary capture area and is deliberately not version-controlled by default (adjust .gitignore yourself if you want to keep them).

Four Entry Points

The same storage is shared by four surfaces, with identical ID scheme and file format:

  • CLI: backlog memo create/list/view/update/delete, the fastest terminal capture path. See CLI Memo Commands.
  • Web: the /memos page stream for quick capture + calendar mode. See Web Memo Page.
  • MCP: memo_create / memo_list / memo_view / memo_update / memo_delete tool group, directly readable and writable by AI agents.
  • HTTP API: /api/memos* routes (list, calendar counts, read/write by ID) for scripts and third-party tools.

In addition, memos are integrated into Global Search; the CLI provides backlog instructions memos usage guide, and the MCP side provides the backlog://workflow/memos workflow resource.

Usage Recommendations

  • Capture lightly: one memo says one thing; write it and forget it, relying on search and links to find it later.
  • Use tags: use --tags or inline #tag in the Web UI to create coarse-grained partitions of the inbox.
  • Write entity IDs: casually include IDs like back-123, doc-5 in the body; they automatically become links when rendered.
  • Digest regularly: memos are an inbox, not a destination; important conclusions should be promoted to tasks or documents. This release does not provide an automatic promotion path; the bridging method is to reference the memo ID in a task body.

results matching ""

    No results matching ""