Pick an option on each card and press Done; your answer goes straight
back to the session. The raiser's pick is marked. "I'm not sure" and "Take it to the engineering bridge" are real answers.
1
In plain terms. When someone asks Agent Smith 'what commit is production on?', should Smith be allowed to answer with the commit code? A new rule stops Smith sprinkling machine codes into messages meant for people, but taken literally it also blocks the one case where the code IS the answer to a direct question. The recommendation is to allow it only when a person explicitly asked, and keep everything else banned.
The full question, as raised
agent-smith#1163 adds a ban on machine identifiers in a reply to a person, with a closed set of named exceptions ('TWO STANDING EXCEPTIONS, and there are no others', prompts.py:221, pinned at tests/test_brevity_prompts.py:124). The List-X rule at :485-486 says to NAME the actual items; the seam allows 'an id the reader will type, click or quote back'; the stays-OUT list bans shas; and 'there are no others' overrides the seam. So a direct question — 'what sha is prod on?' — has no legal answer. Two exits: (a) a narrow third allowance for a sha the person explicitly asked for, keeping the closed set; (b) drop 'and there are no others' and let the seam decide case by case.
Why, and what was tried
The session raising this recommends: Allow a commit code when someone asked for it This is the lane's recommendation and I agree with its reasoning. The open seam ('an id the reader will type, click or quote back') is exactly what a model rationalises its way back through, and that rationalising is what produced this problem in the first place — the force of the rule lives in the list being CLOSED. Letting Smith judge each case buys the one 'what sha is prod on?' case and pays for it by reopening every other identifier to the model's own judgement. A narrow 'a sha they asked for by name' costs one clause and leaves nothing else loose. Never showing them is the cheapest option and is defensible, but it means Smith has to refuse or waffle when asked a direct factual question, which reads worse to a person than the noise this rule exists to stop. Everything else in that change is settled and will ship without this; the clause can land as a one-line follow-up either way, so nothing is waiting on the answer. Tried first, unsuccessfully: Not run as a separate round: this is a sub-question of a blocking review finding that the reviewer ITSELF routed to Gera in those words ('That's Gera's call, and one line in the thread is enough'), and it is a preference about how Smith should sound to a person, not a fact a model can establish. The lane verified the mechanical half by RUNNING the detector: session_refs returns () for a bare 8-hex and only matches after the words session or sid.
2d38f47d is parked on this.
2
In plain terms. A finished change is waiting because of one thing nobody has checked: the little Mac that posts meeting transcripts might be missing the address book the change switches over to. If it is missing and we go ahead, that transcript feed stops quietly and nothing alerts anyone. Two quick reads on that machine settle it — or we build it its own address book first, which is more work and pointless if it already has one.
The full question, as raised
agent-smith#1202 is green, reviewed and held: it drops a hardcoded Slack channel id for a registry lookup, but the Granola seat on the Clara mini may run from a checkout that has no registry, which would silently stop the #transcripts feed. Clearing it needs two reads on that mini: does config/automations.toml exist, and does config/channels.toml contain a transcripts row? How should the hold be cleared?
Why, and what was tried
The session raising this recommends: Run the two checks on the Granola mini and report The hold turns on one fact on another machine, and nobody has been asked yet. Tried first, unsuccessfully: fable-decide f4c6b6685 ruled RESOLVED, both rungs independently agreeing at high confidence: the clearing act is a check on another machine, not a judgement. Its reversed pass added that a bare directory listing is not enough — a stale clone whose registry predates the key is the known failure mode, so the transcripts row itself must be grepped. Merging now would breach Constitution section 6 (the replacement lands first); shipping a registry unasked risks a duplicate source of truth if a checkout already exists.
2d38f47d is parked on this.
Answered and withdrawn (3)
b92159301 Should 'main is red' and 'things are down' messages go to #alerts, as Gera said yesterday, or stay in #agent-smith, as Fede ruled two days ago and wrote a test to enforce? This session is going Gera's way meanwhile — those two to #alerts, quieter status notes to Gera's DM, #agent-smith left with the morning and nightly queues plus big Smith updates. Confirming because Gera's way means retiring a rule Fede put in writing with his name on it, and two other sessions are mid-flight proving Fede's direction right now. — answered: Gera's way: red-main and down go to #alerts
b92161877 Agent Smith made four junk posts in #agent-smith today that Gera's ruling says should not be there — a failed-compact notice, a tripwire reporting zero, and two remediation tables. May this session delete them? Separately there are four 'bake alert' posts from 9am that are also off-list by the letter of the ruling, but they are the ONLY record of three live data problems — they were never copied to the alerts channel — so deleting those would lose the findings, and the recommendation is to keep them and fix where they get sent instead. — withdrawn
b92220652 Should the stuck-PR note stay in #agent-smith, or move to Gera's DM? About once a week Smith posts a short list of pull requests still held, where the thing they were waiting on has already happened — finished work needing one click from Gera to release. One landed in #agent-smith this morning. Gera asked that channel to carry only the nightly and morning queues plus big Smith updates. Staying means Gera sees it alongside the queues; moving means it is quieter but easier to miss, and a stuck pull request was already closed unmerged once because nobody saw one of these. — answered: Keep it in #agent-smith