Skip to content

chore: add merge-queue review digest for the open PR backlog - #599

Open
essentialbit wants to merge 2 commits into
mainfrom
chore/merge-queue-digest-20260818
Open

chore: add merge-queue review digest for the open PR backlog#599
essentialbit wants to merge 2 commits into
mainfrom
chore/merge-queue-digest-20260818

Conversation

@essentialbit

Copy link
Copy Markdown
Owner

Summary

  • New scripts/merge_queue_digest.py (read-only, no git/GitHub writes) buckets the open, non-draft PR backlog into a review order: fast chore/fix merges first, small features next, then a needs_attention bucket for large diffs (250+ lines) or a title-level risk-keyword hit, with everything else in other.
  • Targets the real standing bottleneck documented in project memory: pr_merge_planner.py already shows the 33-PR backlog is mergeable and conflict-free, but there's been no merge to main since PR AI Universe tab: live prices, sentiment badges, nav-highlight fix, sector scope #549 (2026-08-03) because a live user session has to work through the whole backlog with no sense of which PRs are quick reviews vs which need real attention.
  • Deliberately scans only the title with word-boundary matching for the risk-keyword check, not risk_rules.classify_risk()'s full-body substring match — verified before shipping that a full-body scan false-positived on 10/33 real PRs in this backlog (e.g. "auth" matching "author", "token" matching "tokenizer", "session" matching "Claude Code session"). Title-only + word-boundary dropped that to 1/33 on the same data.
  • Also reports touches_high_risk_file per PR (main.py/memory_store.py/config.py/crypto_utils.py/CI workflows) as an informational flag, not a bucket driver — folding it into risk scoring would swamp every bucket since nearly every feat: PR wires a new route into main.py.

Test plan

  • PYTHONPATH=. venv/bin/python3 scripts/merge_queue_digest.py runs clean against the live 33-PR backlog, produces valid JSON with sane bucket counts (17/4/2/10)
  • venv/bin/python3 -c "from main import *" still imports cleanly
  • Manually verified the false-positive fix by diffing keyword hits before/after (10/33 → 1/33)

🤖 Generated with Claude Code

Saifodius and others added 2 commits August 18, 2026 00:04
pr_merge_planner.py already confirms the 33-PR backlog is mergeable/
conflict-free; the real bottleneck (fredai.md open item 2) is a human
working through it one PR at a time with no sense of review order.
Buckets by conventional-commit kind + diff size into a fast-to-clear
queue, with a separate needs_attention bucket for large diffs or a
title-level risk-keyword hit (word-boundary matched, title only --
verified a full-body substring scan false-positived on 10/33 PRs
before shipping, e.g. "author"/"tokenizer"/"Claude Code session"
matching auth/token/session).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant