weekly sync — 2026-05-25#5
Open
norbert-pap wants to merge 1 commit into
Open
Conversation
Port the pre-emit verification gate from garrytan/gstack#1642 (v1.43.2.0) into code-review and eng-review skills. Every finding must now quote the specific motivating code line (file:line + verbatim text) before being promoted; unverified findings are downgraded to Nit/FYI (code-review) or dropped silently (eng-review). Includes framework-meta nudge for ORM and metaclass-generated symbols. Also adds .sync-state.json recording this sync run.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Applied changes
1. Pre-emit verification gate →
code-review+eng-reviewSource: garrytan/gstack#1642 (v1.43.2.0 — "fix wave: post-Daegu paper-cut")
Why cherry-picked: Kills an entire class of false-positive findings ("field doesn't exist on model X") by requiring every finding to cite a specific code line before it's surfaced. Self-contained rule with no gstack-internal dependencies.
code-review/SKILL.md— Added### Pre-emit verification gatesection inside Phase 2 (Five-Axis Review), before Axis 1. Rule: quote file:line + verbatim text that motivates the finding; unverified findings are downgraded to Nit/FYI (suppressed from Critical/High). Includes framework-meta nudge for ORM/metaclass symbols (DjangoMeta, Railshas_many, SQLAlchemyrelationship, TypeORM decorators, Prisma generated client).eng-review/SKILL.md— Added## Pre-emit verification gatesection before## CRITICAL RULE. Same quotation requirement; unverified findings are dropped silently rather than presented via AskUserQuestion.Porting notes: Dropped the internal
#1539issue reference. Adapted the confidence-score suppression (gstack uses 1–10 numeric scale) to claudebert's severity labels (Critical/High/Nit/FYI for code-review; drop-silently for eng-review). Core rules and framework-meta nudge kept verbatim.Deferred / noteworthy (not applied)
anthropic-skills PR #1164 — claude-api skill updates
New managed-agents docs:
managed-agents-self-hosted-sandboxes.md,EnvironmentWorkerclass, mid-sessionsessions.update()for tool/MCP overrides,GET /v1/environments/{id}/work/stats. No mapping in claudebert's pipeline (no claude-api skill exists). Would require creating a new skill — out of scope for a cherry-pick sync.gstack PR #1678 — v1.44.0.0 long-lived sidebar
PTY keepalive (25s ping/pong), dual session/auth token model, ring-buffer scrollback replay, terminal-agent watchdog,
--superviseCLI flag. Pure gstack PTY server infrastructure — noSKILL.mdchanges. Not applicable.gstack PR #1645 — v1.43.3.0 fix(browse): headed-mode idle timer
browse/src/server.tsbug fix: four lifecycle handlers were reading a stale module-levelBrowserManagerinstead of the embedder-supplied one. Implementation fix only — noSKILL.mdchanges. Not applicable.VoltAgent/awesome-agent-skills — new skill additions
Three new skills added this week:
digital-marketing-pro,career-ops,dembrandt-skills. Repo has 22,976 stars (well above 50-star threshold). None fill claudebert pipeline gaps (review, qa, design-review, browse, code-review, eng-review, prod-review, design-doc).Sources checked (6)
Generated by Claude Code