Make the consult habits binding, so the label stops drifting by tomorrow #438

New issue

Closed

opened 2026-08-13 12:55:37 +00:00 by coilyco-ops · 0 comments

coilyco-ops commented

2026-08-13 12:55:37 +00:00

Member

Slice of #437, where Darren measured the drift and corrected it by hand, then said plainly:

The queue is currently honest. It will drift again by tomorrow without one of the fixes above, because nothing structural changed.

This is the structural half.

Scope

Both habits go into AGENTS.md, which is what makes them binding on every agent working this repository rather than advice in a thread that scrolls away:

recording a decision removes consult in the same call
asking a human a question adds it in the same call

Plus a doc holding the measurement, so the next person to wonder whether this matters reads a number rather than an assertion: the queue advertised eighteen items when eight were real, while hiding five.

Why habits rather than a check

A check has to decide whether a comment contains an unanswered question, which is a judgement rather than a pattern. Both directions are easy to detect approximately and hard to detect correctly, and a hygiene check that is wrong sometimes trains people to ignore it — which is the same failure as a permanently red security row.

Darren's flag-rather-than-enforce sweep is the reasonable middle and is recorded in the doc as worth building when someone wants it. It is not a prerequisite.

Not in scope

Re-correcting the labels. Darren already did both directions by hand and the queue is honest as of filing.

Acceptance

Both habits are in AGENTS.md under the agent rules that already bind PR bodies and prose style.
The doc records the measured cost, not just the rule.

**Slice of** https://forgejo.coilysiren.me/coilyco-gaming/sirens-echo/issues/437, where Darren measured the drift and corrected it by hand, then said plainly: > The queue is currently honest. **It will drift again by tomorrow** without one of the fixes above, because nothing structural changed. This is the structural half. ## Scope Both habits go into `AGENTS.md`, which is what makes them binding on every agent working this repository rather than advice in a thread that scrolls away: - recording a decision **removes** `consult` in the same call - asking a human a question **adds** it in the same call Plus a doc holding the measurement, so the next person to wonder whether this matters reads a number rather than an assertion: the queue advertised eighteen items when eight were real, while hiding five. ## Why habits rather than a check A check has to decide whether a comment contains an *unanswered* question, which is a judgement rather than a pattern. Both directions are easy to detect approximately and hard to detect correctly, and a hygiene check that is wrong sometimes trains people to ignore it — which is the same failure as a permanently red security row. Darren's flag-rather-than-enforce sweep is the reasonable middle and is recorded in the doc as worth building when someone wants it. It is not a prerequisite. ## Not in scope Re-correcting the labels. Darren already did both directions by hand and the queue is honest as of filing. ## Acceptance - Both habits are in `AGENTS.md` under the agent rules that already bind PR bodies and prose style. - The doc records the measured cost, not just the rule.

coilyco-ops referenced this issue from a commit

2026-08-13 12:55:47 +00:00

docs(consult): make the label track the thread, since nothing else does

coilyco-ops referenced this issue from a pull request that will close it,

2026-08-13 12:56:15 +00:00

docs(consult): make the label track the thread, since nothing else does #439

coilyco-ops referenced this issue

2026-08-13 12:56:20 +00:00

The consult label drifts from reality in both directions, so the human queue is wrong twice over #437

coilyco-ops closed this issue

2026-08-13 12:58:47 +00:00