Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions docs/hygiene-history/ticks/2026/05/01/2105Z.md
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
| 2026-05-01T21:05:00Z | opus-4-7 / autonomous-loop tick | 98fc7424 | Review-fix + new-bullet tick on PR #1160 (meta-rule). **REFRESH WORLD MODEL**: 7 PRs (4 merged, 3 open). #1160 had 3 unresolved review threads β€” investigated per the BLOCKED-with-green-CI wake-time discipline. **Three findings + one new framing landed**: (1) Codex P2 + Copilot P0 dup β€” SQLSharp DI pattern bullet pointed at memory file that's on PR #1155's branch (not main); CLAUDE.md pointer would dangle when #1160 lands. Fix: remove the SQLSharp bullet entirely from #1160; defer to follow-up after #1155 merges. Each PR's pointer targets must exist on main at merge time. (2) Copilot β€” ".claude/rules" quoted from maintainer's verbatim message; Zeta doesn't have that directory (uses skills/agents/commands). Fix: brief inline note clarifying `.claude/rules/` IS canonical Anthropic per `code.claude.com/docs/en/memory` but Zeta's discoverable surfaces are `.claude/{skills,agents,commands}/`. Adopting `.claude/rules/` is a viable future addition. Per maintainer 2026-05-01 ("we can do whatever here to get it right IDK the right answer it's probably play around a bit") the clarification stays open-ended rather than locking in a position. (3) New maintainer framing 2026-05-01: *"it could just remind to you use the router as lookup of existing substrate, quick inventory via router"* + *"the claude.md should be sufficient for him to figure out skill router as substrate inventory"* (buddy-test design constraint: no priming). Added new CLAUDE.md ground-rule bullet β€” **skill router as substrate inventory before authoring**. Before writing new memory file / rule / skill / agent / doctrine, search the available-skills list AND `.claude/{skills,agents,commands,rules}/` on disk for substrate on the same topic. The router's description-keyed search IS Zeta's structured-substrate index; using it as inventory pre-authoring prevents goldfish-ontology recurrence. The bullet IS the wake-time encoding so fresh-instance buddies inherit the discipline without primed prompting. All 3 threads resolved. Pointer-target verification done before commit (only existing-on-main + landing-in-this-PR memory files referenced). Cron 98fc7424 healthy. | [PR #1160 +1 commit (34e7c78); 3 review threads resolved; new skill-router-as-substrate-inventory bullet β€” landed at wake-time scope so buddy-test (and normal Otto) inherit the discipline] | The buddy-test design constraint (no primed prompting; CLAUDE.md must be sufficient for fresh instance to discover the discipline) is itself a substrate-design principle: any CLAUDE.md bullet must teach future-Otto how to find substrate, not just contain the substrate. Today's worked example: instead of the meta-rule alone telling future-Otto "rules must land in CLAUDE.md or pointer," the inventory-via-router bullet teaches future-Otto HOW to discover existing substrate (via skill router + .claude/ directory listing) before authoring new substrate. Two-bullet composition: meta-rule says where lessons must land; inventory-router bullet says how to find existing lessons before writing. Maintainer's "i got an idea forming" + "play around a bit" framing β€” the .claude/rules adoption question stays open as exploratory. |
Loading