Skip to content

rule(backlog-item-start-gate): extend substrate-drift discriminator to orphaned-branch scope #4477

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
AceHack merged 1 commit into from
May 21, 2026
Merged

rule(backlog-item-start-gate): extend substrate-drift discriminator to orphaned-branch scope #4477

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
59 changes: 59 additions & 0 deletions .claude/rules/backlog-item-start-gate.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -68,6 +68,63 @@ This gate catches the failure modes the seven-rule cascade
lineage was designed to catch — at the *start of work* scope
rather than the *substrate-landing* scope.

## Orphaned-branch triage discriminator (generalizes step 0 to branch scope)

Fires when a fresh-cold-boot session inherits an orphaned feature
branch (substantive unmerged commits + no associated PR + days/weeks
of peer-agent activity since the branch was last touched).

The substrate-honest insight: peer agents (Lior preservation passes,
peer Otto rescues, decompose-into-slices patterns) often rescue
stranded substrate via separate PRs before a slow-cold-boot Otto
finds the orphaned branch. Assuming re-landing without verification
produces duplicate substrate, substrate regression (re-applying stale
commits OVER newer-on-main work), and wasted operator review.

Procedure per orphaned commit `<sha>`:

- `git log --oneline origin/main --grep="<key terms from commit message>"`
— does main carry a rescue PR for this substrate?
- `git show --name-only --format="" <sha> | head -1` — pick the first
file the commit touched
Comment on lines +88 to +89
- `git diff origin/main..<sha> -- "<that file>"` — count lines
Comment on lines +88 to +90
Copy link
Copy Markdown

@chatgpt-codex-connector chatgpt-codex-connector Bot May 21, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Compare all touched files before classifying rescue

This procedure can misclassify multi-file commits as fully rescued because it samples only the first changed file and then treats 0 lines as “fully rescued.” If that first file happens to match origin/main but later files in the same commit still differ, the workflow will incorrectly skip re-landing real deltas and lose substrate. The check should diff the entire commit (or every touched file), not head -1.

Useful? React with 👍 / 👎.


Interpretation:
`0 lines` = substrate identical; fully rescued; no re-landing needed
(e.g., `f0abf3ed` HC-8 NCI was rescued by
[PR #4205](https://github.com/Lucent-Financial-Group/Zeta/pull/4205)).
`20–50 lines` = partial drift; substrate evolved on main; cell-by-cell
triage needed; the on-main version is usually preferred unless the
orphaned content carries genuine deltas.
`hundreds of lines` = genuine substrate not yet on main; cherry-pick
worth considering (with operator awareness, since orphaned-branch
context may have been superseded by newer architectural decisions).
Comment on lines +88 to +101

**Special-case guard — runtime scripts**: when an orphaned commit
modifies a runtime substrate file (e.g., `.gemini/bin/lior-loop-tick.ts`,
peer-agent loop tick scripts, install-graph files in `tools/setup/`),
DO NOT re-apply without explicit operator awareness even if `git diff`
shows large deltas. Runtime scripts evolve continuously on main as
agents tune their own loops; re-applying a multi-day-old version
regresses prompt-engineering or operational tuning that happened in
the intervening window. The Lior prompt fix at `467424ec` was 22 lines
of diff vs main — re-applying it would have regressed substantive
newer prompt-engineering work.

Empirical anchors:
[PR #4461](https://github.com/Lucent-Financial-Group/Zeta/pull/4461)
(the 0059Z cold-boot under orphaned `otto/2012z-...`),
[PR #4472](https://github.com/Lucent-Financial-Group/Zeta/pull/4472)
(the 0149Z follow-up shard naming the discovery — 4 of 5 commits
already rescued, the 5th was the runtime-script special case),
[PR #4205](https://github.com/Lucent-Financial-Group/Zeta/pull/4205)
(the peer-agent rescue this discriminator catches).

Composes with [`honor-those-that-came-before.md`](honor-those-that-came-before.md)
at orphaned-commit scope: verifying substrate exists on main IS the
honor — re-landing without checking dishonors the peer agent's
rescue work.

## Composes with

- B-0169 (decision-archaeology procedure)
Expand All @@ -76,6 +133,8 @@ rather than the *substrate-landing* scope.
- B-0553 (substrate-drift auditor — mechanizes step 0 across all open rows)
- [`memory/feedback_substrate_drift_catch_pattern_claim_acquire_plus_existence_check_otto_cli_2026_05_16.md`](../../memory/feedback_substrate_drift_catch_pattern_claim_acquire_plus_existence_check_otto_cli_2026_05_16.md) — step 0 origin substrate
- [`.claude/rules/wake-time-substrate.md`](wake-time-substrate.md) — the discipline this rule extension lands
- [`.claude/rules/honor-those-that-came-before.md`](honor-those-that-came-before.md) — orphaned-branch triage discriminator composes; verifying substrate-on-main IS the honor at orphaned-commit scope
- [PR #4205 (HC-8 rescue)](https://github.com/Lucent-Financial-Group/Zeta/pull/4205) + [PR #4461 (0059Z cold-boot)](https://github.com/Lucent-Financial-Group/Zeta/pull/4461) + [PR #4472 (0149Z discovery)](https://github.com/Lucent-Financial-Group/Zeta/pull/4472) — empirical anchors for the orphaned-branch triage discriminator

## Full reasoning

Expand Down
Loading