pr-review-considerations
Skill742 starsChanged 4 months agoMITView on GitHub
Additional iTwin.js PR review heuristics focused on consumer impact, compatibility, invariants, UI evidence, and docs accuracy.
In this repository6 other files it gives its agents.
- AGENTS.mdAGENTS.md
- api-promotionSkill
- backport-resolutionSkill
- cve-remediationSkill
- merge-conflict-resolvingSkill
- test-suite-refactoringSkill
The file
We split this skill into the rules it gives an agent
10 rules. About 0 are common, 9 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Use this skill when reviewing pull requests, preparing review guidance, or deciding what GitHub Copilot should pay attention to in this …
- Distinctive: line 18
Prefer comments that explain *why* the issue matters to consumers, downstream apps, or maintainers.
- Distinctive: line 26
If a change is declared safe because of fallback behavior, verify the fallback exists and ask for the exact helper or test when the proof …
- Distinctive: line 28
When integration risk exists, ask whether the change was validated in downstream apps, learning snippets, or consumer patch testing.
- Distinctive: line 33
Remove redundant or duplicated state when a single source of truth is available.
- Distinctive: line 35
Prefer extracting shared helpers over duplicating similar conditionals, heuristics, or parsing logic.
- Distinctive: line 36
Call out ad hoc string-matching heuristics or repeated condition fragments that should be centralized.
- Distinctive: line 41
Check CTA placement, layout shift, copy clarity, localization wording, accessibility, and whether UX feedback is warranted.
- Distinctive: line 47
Prefer compiled or extracted snippets over handwritten README code that can drift.
- Very specific: line 49
Only ask for docs/changehistory/NextVersion.md and migration guidance when the change is breaking or is a significant new user-visible …
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
PR Review Considerations
Use this skill when reviewing pull requests, preparing review guidance, or deciding what GitHub Copilot should pay attention to in this repository.
GitHub Copilot PR review on GitHub.com does not load repository skills directly. To influence GitHub review comments, keep the concise version of this guidance in
.github/instructions/pr-review.instructions.md. GitHub's current repository instruction model also means that file can influence other Copilot interactions for matching files, not only PR review.
Review stance
These considerations supplement — not replace — standard code review practices. Continue checking for correctness, security, performance, and readability as you normally would. The guidance below adds iTwin.js-specific domain concerns on top of that baseline.
- In addition to standard review checks, prioritize issues likely to cause runtime regressions, consumer confusion, integration churn, or repeated reviewer back-and-forth.
- Skip pure style-only nits unless they hide a correctness problem or conflict with an established repo pattern.
- Prefer comments that explain why the issue matters to consumers, downstream apps, or maintainers.
What reviewers repeatedly look for
1. Consumer impact and compatibility
- For
package.json, peer dependency, or version bumps, ask whether each bump is necessary and whether the peer range can stay broader. - Watch for unrelated dependency churn getting folded into a feature PR.
- If a change is declared safe because of fallback behavior, verify the fallback exists and ask for the exact helper or test when the proof is missing.
- For parser, persistence, schema, or serialization changes, check old-data compatibility and roundtrip behavior.
- When integration risk exists, ask whether the change was validated in downstream apps, learning snippets, or consumer patch testing.
2. Stronger invariants and less ambiguity
- Push
undefinedand loading handling upward instead of spreading optional checks through the tree. - Remove redundant or duplicated state when a single source of truth is available.
- Encode behavioral requirements in types when possible, especially prop combinations tied to modes like
editable. - Prefer extracting shared helpers over duplicating similar conditionals, heuristics, or parsing logic.
- Call out ad hoc string-matching heuristics or repeated condition fragments that should be centralized.
3. UI and UX expectations
- UI workflow changes should include a GIF or screenshot in the PR description.
- Check CTA placement, layout shift, copy clarity, localization wording, accessibility, and whether UX feedback is warranted.
- Confirm that app-specific glue is not being accidentally presented as reusable package behavior.
4. Docs, examples, and change notes
- README content, comments, screenshots, and example snippets should match what actually ships.
- Prefer compiled or extracted snippets over handwritten README code that can drift.
- For public or beta API changes, verify release tags,
rush extract-api, and requiredrush changefiles. - Only ask for
docs/changehistory/NextVersion.mdand migration guidance when the change is breaking or is a significant new user-visible feature. - In changelog text, describe concrete user-visible behavior instead of labeling every bug fix as "breaking."
High-signal review prompts
- "If this is safe because of a fallback, can you point to the helper or test that proves it?"
- "Can this invariant be expressed in the type instead of repeated runtime checks?"
- "Does this peer dependency bump need to happen in this PR, or can the range stay broader?"
- "Will this behavior survive save/load or serialize/deserialize roundtrips?"
- "Please add a GIF or screenshot so reviewers can validate the UI workflow quickly."
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

