Skip to content

ci(panels): pin target=copilot for pr-review-panel and triage-panel#1192

Merged
danielmeppiel merged 1 commit into
mainfrom
chore/panels-target-copilot
May 7, 2026
Merged

ci(panels): pin target=copilot for pr-review-panel and triage-panel#1192
danielmeppiel merged 1 commit into
mainfrom
chore/panels-target-copilot

Conversation

@danielmeppiel
Copy link
Copy Markdown
Collaborator

TL;DR

Pin target: copilot on both panel workflows now that shared/apm.md exposes the new target: input (#1184).

Why

Both panels run on the gh-aw default Copilot CLI engine. Without an explicit target: they fell back to all, which:

  1. Installed seven harnesses (vscode/claude/cursor/opencode/codex/gemini/windsurf) on every panel run for no benefit -- only .github/copilot-instructions.md is consumed by the engine.
  2. Was the path that crashed the apm self-check (run 25511043293) before fix(install,test): exclude cowork from --target all + stabilize defer-start timer test #1191 landed, because all was leaking the experimental copilot-cowork target into project-scope installs.

What

  • .github/workflows/pr-review-panel.md: add target: copilot to the shared/apm.md import.
  • .github/workflows/triage-panel.md: same.
  • gh aw compile regenerated both lock files; verified the substitution baked through (target: copilot in the lock, frontmatter_hash updated, agent_id: copilot unchanged).

Defence in depth

Even if --target all regressed again upstream, these workflows now request a single supported harness explicitly, so an experimental target leaking into all could not re-break the panels via this code path.

Both panels run on the gh-aw default Copilot CLI engine, so they only
need the 'copilot' layout from the apm package. Pinning target=copilot
(via the new shared/apm.md input from #1184) ships a slim per-harness
bundle instead of installing all seven harnesses on every panel run.

This also side-steps the cowork-in-all crash that hit the apm
self-check (run 25511043293) before #1191 landed; the panels now
explicitly request a single supported harness, so an experimental
target leaking into 'all' could not regress these workflows again.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot AI review requested due to automatic review settings May 7, 2026 18:14
@danielmeppiel danielmeppiel merged commit 83431db into main May 7, 2026
17 checks passed
@danielmeppiel danielmeppiel deleted the chore/panels-target-copilot branch May 7, 2026 18:15
Copy link
Copy Markdown
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

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

Pull request overview

Pins the gh-aw panel workflows to a single APM harness (target: copilot) so the generated bundle only includes the Copilot/GitHub harness layout needed by the panel engine, avoiding unnecessary multi-target packing.

Changes:

  • Set target: copilot on the shared/apm.md import in both panel workflow sources.
  • Regenerated both gh-aw lock workflows so the packed APM bundle uses target: copilot instead of all.
Show a summary per file
File Description
.github/workflows/triage-panel.md Adds target: copilot to the shared/apm.md import inputs.
.github/workflows/triage-panel.lock.yml Regenerated lock workflow; pack step now uses target: copilot.
.github/workflows/pr-review-panel.md Adds target: copilot to the shared/apm.md import inputs.
.github/workflows/pr-review-panel.lock.yml Regenerated lock workflow; pack step now uses target: copilot.

Copilot's findings

  • Files reviewed: 4/4 changed files
  • Comments generated: 0

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.

2 participants