smoke(simplify): first real use — one detached run of a `main` build on two notes-fixture smoke issues, after #S-01 to #S-07 merge; the owner reads the report

Open
  • Is something wrong?
#1,655 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
25/100
Issue type
Feature
Clarity
Clearly specified
Activity status
Active
Tech stack
github-actions, node.js, typescript
Domain
devtools, testing-qa

Research direction

Read the smoke templates and bin/cli.ts:374, then inspect the merged main build and the frozen skill referenced by the report. Follow the procedure to file two smoke issues on admarble/notes-fixture, launch the detached run, and collect stdout, the run log, and phase transcripts. Done when the tracker report includes the acceptance evidence and the owner answers SMOKE: ship or SMOKE: reopen #N <reason>.

Written by the indexing model from the issue text.

Description

graph-2026-11-simplify

Epic: graph-2026-11-simplify Tier: judgment Execution: runner Blocked by: #1646, #1647, #1648, #1649, #1650, #1651, #1652

  • Blocked by: #1646
  • Blocked by: #1647
  • Blocked by: #1648
  • Blocked by: #1649
  • Blocked by: #1650
  • Blocked by: #1651
  • Blocked by: #1652

Problem

The phases that build each node run the published 2.20.0, not the branch, so no node's runtime behaviour has been seen by the time it merges. Unit tests show the halt, the warning and the reach block in isolation; they do not show a live qa agent leaving a needs-human Evidence line unrun while the frozen skill's §6h tells it to run every line (OQ-3), or which verdict follows (OQ-6). This node runs the merged main once. Founding doc P-12.

Procedure

  1. Build main at the commit after the last of #1646 to #1652 merges; record the SHA.
  2. On admarble/notes-fixture, file two smoke issues from main's smoke templates, written at launch: A with three ACs (one ordinary npx vitest AC; one whose Evidence is npx tsx --env-file=.env.local scripts/smoke-check.ts → ok; one whose Evidence cannot pass, so that QA refuses), B with one ordinary AC. Neither touches fixture #34's branch.
  3. Launch node ~/Projects/sequant/dist/bin/cli.js run <B> <A> --chain -Q --max-iterations 3 with launchctl submit, under smpl-sequant/experiments/E2c/deny.sb as it stands (#856; sandbox per P-12). B goes first because --chain halts on any failed issue (bin/cli.ts:374) and A is built to fail.
  4. Collect the stdout, the run log and the phase transcripts; write the report as a comment on the tracker.

Acceptance Criteria

  • AC-1: The run printed both experimental warnings. Evidence: grep -cE '^⚠ (--chain is|Multi-issue runs are) experimental' ~/Projects/notes-fixture-runs/simplify-smoke/stdout.log → ≥ 2
  • AC-2: The run summary named A's env-file AC as needing a person. Evidence: grep -c 'needs a person: .npx tsx --env-file' ~/Projects/notes-fixture-runs/simplify-smoke/stdout.log → 1
  • AC-3: No phase executed the env-file command (the prompt block quoting it is a user turn, not a tool call). Evidence: ls ~/Projects/notes-fixture-runs/simplify-smoke/transcripts/*.jsonl > /dev/null && cat ~/Projects/notes-fixture-runs/simplify-smoke/transcripts/*.jsonl | jq -r 'select(.type=="assistant") | .message.content[]? | select(.type=="tool_use" and .name=="Bash") | .input.command' | grep -c -- '--env-file' → 0
  • AC-4: The run log shows A's quality loop stopping at ac_not_met_repeated, or the report says the second refusal was not reached and why. Evidence: jq -r --argjson a "$A" '.issues[] | select(.issueNumber==$a) | .abortReason // "not-reached"' ~/Projects/notes-fixture-runs/simplify-smoke/run-log.json | cut -d: -f1 → ac_not_met_repeated, or not-reached with a reason in the report
  • AC-5: The removed flags are rejected by the build. Evidence: for f in --model-ladder=sonnet --escalate-effort --stacked --reflect --full-qa; do node ~/Projects/sequant/dist/bin/cli.js run 1 $f --dry-run 2>&1 | grep -c "unknown option"; done | sort -u | tr '\n' ' ' → 1
  • AC-6: The owner answered the smoke question on the tracker. Evidence: gh issue view $(gh issue list --label graph-2026-11-simplify --search 'tracker' --state all --json number --jq '.[0].number') --json comments --jq '[.comments[].body|select(test("^SMOKE: (ship|reopen)"))]|length' → ≥ 1

The question (owner): given the report, is main fit to release as 3.0, or which node reopens? Answer SMOKE: ship or SMOKE: reopen #N <reason>.

isGateTestEvidence classifies AC-1 to AC-4 as gates (parse-check, 2026-11-04). This node changes nothing in sequant's tree and runs no /qa phase (Execution: runner), so §6i never reads them and no SEQUANT_MUTATION record applies; the report pastes each command's output.

OQ-6 (the verdict A gets with one AC pending) is recorded in the report; a verdict the frozen skill gets wrong is a post-lift fix, not a reopen.

May touch

Nothing in sequant's tree. admarble/notes-fixture: two issues, their branches and PRs, closed after the report. ~/Projects/notes-fixture-runs/simplify-smoke/ (local).

Non-Goals

  • Any scored measurement; this is a smoke.
  • Fixture #34 or any fixture with ground truth.

Done when

The owner has read one real run of the simplified main and answered ship or reopen.

Dominant language
TypeScript
Stars
1
Forks
1
Avg merge
11h 3m
Merged PRs (30d)
84

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from sequant-io/sequant

All issues in sequant-io/sequant

Similar issues

More TypeScript issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.