Repository navigation
feat(starter): include Pick & polish in feedback reviews - #78
Merged
Merged
Conversation
lukasoppermann
added this pull request to stack #43
October 22, 2026 14:14
lukasoppermann
marked this pull request as draft
October 22, 2026 14:20
lukasoppermann
changed the title
lukasoppermann
marked this pull request as ready for review
October 22, 2026 14:24
There was a problem hiding this comment.
🔵 Needs a closer look
Fresh-host validation of mixed-source review, private handoff and approval-gated application remains incomplete.
No open issues 🎉
What changed in this PR
Extends Starter’s feedback workflow to include Hub-synchronized Pick & polish conversations and carry selected findings into approval-gated guidance proposals. Runtime APIs, UI, publishing and automatic Hub Learnings remain unchanged.
Changes:
- Collect complete mixed-source conversations while distinguishing human evidence from agent context.
- Define a private version 2 handoff, retain review-only version 1 compatibility, and require explicit approval before guidance edits.
- Add metadata documentation, synthetic evaluation scenarios and proposal-skill copy checks.
| File | Description |
|---|---|
starter/README.md |
Explains the review-to-proposal workflow and evidence limits. |
starter/.agents/skills/review-hub-feedback/SKILL.md |
Adds mixed-source collection, judging rules and proposal handoff. |
starter/.agents/skills/review-hub-feedback/references/sources.md |
Defines source identities, conversation handling and metadata limits. |
starter/.agents/skills/review-hub-feedback/references/handoff.md |
Defines private evidence records and legacy compatibility. |
starter/.agents/skills/prepare-feedback-proposals/SKILL.md |
Introduces bounded investigation and approval-gated proposals. |
starter/.agents/skills/prepare-feedback-proposals/references/apply.md |
Limits application to explicitly approved guidance changes. |
scripts/fixtures/feedback-review-scenarios.json |
Adds synthetic scenarios with expected review and handoff decisions. |
scripts/create-prototype-repo.test.mjs |
Includes the proposal skill in starter-copy checks. |
Review effort: Max
lukasoppermann
force-pushed
the
lukasoppermann-pick-and-polish-learnings
branch
from
October 22, 2026 15:30
ada6a2f to
7a03fa4
Compare
nshki
approved these changes
Oct 23, 2026
Base automatically changed from
lukasoppermann-canvas-improvement-filters
to
main
October 24, 2026 05:44
Collect complete mixed-source conversations, preserve typed identities and human/agent provenance, and carry evidence into approval-gated proposals. Document storage and API parity limits without changing Hub automatic Learnings. Closes #76 Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1e6c938b-f09a-4ade-88d9-68fed03eb710
Remove the proposal prerequisite, handoff format, synthetic fixture and extra documentation from this PR. Keep only API inclusion, complete conversation retrieval and the source identity and evidence distinctions required by Pick & polish. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1e6c938b-f09a-4ade-88d9-68fed03eb710
lukasoppermann
force-pushed
the
lukasoppermann-pick-and-polish-learnings
branch
from
October 24, 2026 05:44
7a03fa4 to
0da0d41
Compare
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
Minimize and redact forwarded evidence, validate structured inputs, and preserve the raw-retrieval trust boundary without claiming prompt-injection immunity. Restore proposal files removed during the server-side stack rebase, retain typed IDs in the existing handoff, and regenerate plugin skill copies to fix asset drift. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: 1e6c938b-f09a-4ade-88d9-68fed03eb710
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.
Why
The feedback review skill currently starts with summaries that omit Pick & polish. Human requests and corrections from those conversations should be reviewed alongside ordinary comments. Addresses #76 for feedback collection; proposal-flow changes are outside this narrowed PR.
What changes
Changes only
starter/.agents/skills/review-hub-feedback/SKILL.md, stacked above #71. The skill uses the existing full-detail API withimprovementNotes: "include"and follows every conversation's message pages. It preserves distinct review and improvement IDs, the original Build and target, and human-versus-agent context. Human requests use the existing relevance and transfer standards without extra weight; agent answers are context, not independent support or verified fixes. Missing metadata stays unknown.No proposal skills, handoff formats, fixtures, runtime APIs or automatic Hub Learnings changes are included. Fresh-host mixed-source skill execution remains unverified.