note.md + OpenClaw

Why this pairing works

OpenClaw keeps memory in markdown — MEMORY.md for long-term facts and memory/YYYY-MM-DD.md for daily working notes. note.md also uses files, but its wikipage/ and dailynote/yyyy/*.note.md layouts have different paths and schemas. Connect them through an explicit workspace or conversion rule rather than treating the formats as identical.

Pair them and each side gets what it lacks: OpenClaw gets a human who actually reads and curates its memory in a view built for that; you get an agent that works around the clock and writes everything down where you can see it.

Setup

1. Put an AGENTS.md at your vault root describing the conventions (sidecar pairing, daily-note paths, [[yyyy-MM-dd]] date links). Grab the summary from llms-full.txt.
2. Point OpenClaw's workspace at the vault, or have it write reports to a dedicated vault folder. Do not symlink memory/ to dailynote/ without a converter: the layouts and metadata differ.
3. Optionally install the official OpenClaw Chat plugin for a note.md-native conversation window.
4. Have OpenClaw write reports and research as .md documents into the vault.
5. Open them in note.md, read, highlight, question — your marks land in sidecar .note.md files.
6. Tell OpenClaw to read sidecars before follow-up work. Your judgment becomes its steering signal.

The loop in practice

Evening: OpenClaw researches a topic and drops research/topic.md in the vault. Morning: you read it in note.md over coffee, highlight two claims, add a doubt. Afternoon: OpenClaw picks up research/topic.note.md, sees exactly which claims earned your attention, and digs where you doubted. No prompt engineering — just files.

FAQ

Does OpenClaw need a plugin to work with note.md?

No for file-based work: both sides can use plain markdown with an AGENTS.md describing the boundaries. The optional official OpenClaw Chat plugin adds a note.md-native conversation window.

Is it safe to let OpenClaw write into my vault?

Keep the vault in git (see the GitHub guide) so committed agent writes are diffable and revertible. By convention agents treat .note.md sidecars as human-owned; Smart Lookup's fenced answer writeback is the narrow built-in exception. State the boundary in AGENTS.md.