kyleandClaude Opus 5.5 5a0cee42ec docs: M5c spec: conclusions go to <id>_memory; no retraction, notes cannot be undone ...

A memory write is made during a turn, before that turn's Honcho session exists, and a session
whose turns are never synced has none, so each Boxmaker session's conclusions go to its own
<id>_memory session, holding the agent's peer and the owner's. `bxctl memory undo` refuses notes
(they are append-only), so the undo-of-a-fleet-note row and its retraction go. Owner, 2026-09-28.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

2026-09-28 08:53:56 -07:00

..

2026-09-17-m2a-inference-path.md

Add M2a plan: thirteen tasks, tests, fake server and recordings

2026-09-17 13:34:11 -07:00

2026-09-17-pre-m1-design.md

Review M1: accept with two follow-up tasks

2026-09-17 09:14:29 -07:00

2026-09-18-m2b-agent-loop.md

Review M2b: accept with one follow-up task; record lessons

2026-09-18 20:47:03 -07:00

2026-09-18-m3a-decision-path.md

Record the second review of task 23 and the findings first left out

2026-09-22 21:54:28 -07:00

2026-09-22-m3b-runner.md

M3b review fixes by the design model: curl --globoff, podman --pull=never

2026-09-23 15:25:51 -07:00

2026-09-23-m4a-gateway.md

M4a spec: catch-up, recording and state writes, as settled while planning

2026-09-23 19:05:44 -07:00

2026-09-24-m4b-approvals.md

M4b spec: approvals from non-approvers in direct messages go to an approver, with who asked and what

2026-09-24 11:09:59 -07:00

2026-09-24-m5a-local-memory.md

M5a spec: the session's state follows a result's own label; plainer note answers

2026-09-25 17:44:35 -07:00

2026-09-25-roncho-requirements.md

docs: roncho requirements: name no other agent's workspace

2026-09-28 00:38:43 -07:00

2026-09-26-m5b-hardening.md

loopd: the tag check asks /tokenize twice, not once per tag

2026-09-27 19:08:42 -07:00

2026-09-27-m5c-honcho.md

docs: M5c spec: conclusions go to <id>_memory; no retraction, notes cannot be undone

2026-09-28 08:53:56 -07:00