Skip to content

deploy(vera): github-plugin v0.7.0 → v0.7.1, pr-reviewer v0.38.0 → v0.39.0 - #54

Merged
mabry1985 merged 1 commit into
mainfrom
deploy/vera-plugin-pins-0.39
Sep 14, 2026
Merged

mabry1985 merged 1 commit into
mainfrom
deploy/vera-plugin-pins-0.39

Conversation

@mabry1985

Copy link
Copy Markdown
Member

Bumps Vera's plugin pins to the releases that stop her reviewing blind:

The two ship together. v0.39.0's finder prompts pass ref= on every file read, and that only works with v0.7.1's GET fix.

This bumps all three places the pins live, as #51 did: protoagent.bundle.yaml (which docker-publish reads), the Dockerfile's ARG defaults, and the README table (which CI asserts). Core stays at its current version.

After merge, docker-publish builds vera:latest. The deploy is then docker compose pull vera && docker compose up -d vera in homelab-iac/stacks/vera on ava.

🤖 Generated with Claude Code

https://claude.ai/code/session_01LSRkzcPyaDrKSY7geFuGr3

….39.0

Vera's lanes have been reviewing blind for weeks. These two releases fix it:
- github-plugin v0.7.1: the contents reads are GETs (#29). `-f ref=` made gh send
  POST, a 404 on /contents, so every ref-pinned file read failed
  (pr-reviewer-plugin#118).
- pr-reviewer v0.39.0: a blind or broken finder can't read as a clean pass (#120, #122);
  an absent findings payload never posts PASS; a degraded lane caps PASS at WARN; the
  structural lane's turn budget survives the host's step accounting (#121, #119).
They ship together: v0.39.0's finder prompts pass `ref=` on every read, which only
works with v0.7.1's GET fix.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LSRkzcPyaDrKSY7geFuGr3

@protoreview protoreview Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

QA panel review — PASS

code-review-structural · head 34863b9cfe1f · formal

No findings were raised by any review-finder on this PR, and the structural verification pass confirmed an empty array — there is nothing to verify, re-rank, or carry forward. Overall risk is low on the evidence available, but this is a clean bill of health by absence rather than by positive confirmation: the panel found no defects, not that it confirmed correctness. No fix-first item. No panel disagreement. No prior requests to disposition. Gap: the structural pass (code-review-structural:verify) ran as a no-op because it received an empty findings list; it did not independently inspect the diff for structural issues (e.g., missing type annotations, unhandled exception paths, or API contract violations) — it can only verify what the finders flag. If this PR touches public API surfaces or concurrency-sensitive code, consider a targeted structural pass with explicit scope.

No findings — the review came back clean.

findings JSON (machine-readable)
[]

@mabry1985
mabry1985 merged commit 7f1ee37 into main Sep 14, 2026
2 checks passed
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