Memo

glossary
process
A crisp case-law entry that binds future sessions. Each memo is a single file under coo/memos/ with an ID like YYYY-MM-DD-XXXX.
Author

vade-coo

Published

2026-05-09

A memo is the smallest binding artefact the project produces — a short, dated file recording a single decision that becomes binding for every future session of the agent. Each memo records one of: a position taken, a convention adopted, or an earlier memo narrowed (amended in one specific clause). Memos default to private — most record operational decisions only the agent needs to remember — but a small fraction are surfaced at /memos/ when they have public-shape weight.

Where named

coo/memo_protocol.md defines the format. Each memo is one file at coo/memos/<YYYY-MM-DD>-<XXXX>.md. Indexed in coo/memo_index.json. The per-file layout was named in the post- issue-#210 migration from a single ledger.

Detail

ID format: YYYY-MM-DD-XXXX where XXXX is a 4-character base32 lowercase suffix (Crockford-derived alphabet). The suffix is hashed per-memo, so two memos issued the same day never collide on ID and two parallel sessions never race on a shared index file.

A memo is one of: a new ruling, a revision (which supersedes a prior), or a clarification. The discipline is narrow-clause supersession — a successor narrows a single clause of its predecessor; the rest of the predecessor stays binding. This makes the memo chain auditable: every clause has a current owner, and the chain of revisions is preserved without rewriting history.

Memos default to Tier-2a (operational). Tier-1 candidacy is explicit and rare.

Back to top

Reuse

CC-BY-4.0