Conversation
…y extraction Generated from analysis of 695 review comments (2,102 effective weighted) across 4 expert reviewers spanning 10 years of dotnet/fsharp and fsharp/fslang-design review history. Artifacts: - .github/agents/expert-reviewer.md: 15-dimension review agent with 5-wave workflow and folder hotspot mapping - .github/skills/expert-review/SKILL.md: Slim invocation pointer with dimension selection guide and self-review checklist - .github/instructions/ExpertReview.instructions.md: Top 10 compiler review rules scoped to src/Compiler/** - .github/agents/extraction-pipeline.md: Reusable pipeline agent for future reviewer expertise extraction Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Skill name: gerund form (reviewing-compiler-prs) - Description: third person, keyword-rich, specific triggers - Removed redundant prose (Claude already knows) - Added cross-references to complementary skills (review-council, hypothesis-driven-debugging, ilverify-failure, vsintegration-ide-debugging) - Slim pointer pattern: skill is overview, agent is source of truth Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
The review-council skill (generic multi-model review) and expert-review skill (F#-specific dimensions) had 0.55 trigger overlap. Merged into a single reviewing-compiler-prs skill that combines: - F#-specific 15 dimensions with file-based selection - Multi-model 3-phase methodology (briefing, parallel assessment, consolidation with dedup/false-positive filtering) - Self-review checklist Deleted .github/skills/review-council/ — its methodology is now embedded in the merged skill's Multi-Model Review Methodology section. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Description: natural third-person prose, no keyword spam list - Removed redundant opening paragraph (restated description) - Trimmed Multi-Model section: removed briefing pack steps Claude already knows; kept only unique content (assessment gates, consolidation rules) - Removed redundant 'Before requesting...' preamble - 75 -> 54 lines (28% reduction, no content loss) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Agent (expert-reviewer.md): - Dim 1: test layer selection guide (Typecheck/SyntaxTree/EmittedIL/ compileAndRun/Service/Core), happy/negative/interaction paths, assertion quality - Dim 12: pipelines over nesting, active patterns for complex conditions, nesting depth limit (2 levels), wide over deep - Dim 13: ad-hoc if/then band-aid detection, higher-order pattern extraction, search for existing utilities before reimplementing, respect intentional deviations in isolated projects Skill (SKILL.md): - Claims coverage check (orphan claims/changes/partial impls) - 'Correct convention' false positive filter - 'Unexplained != wrong' — doc gap not defect - 'Agents love nitpicks' — deprioritize - Minimum viable council = 2 models Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
… from instructions Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Contributor
✅ No release notes required |
majocha
reviewed
Mar 18, 2026
|
|
||
| --- | ||
|
|
||
| ### 8. IDE Performance & Reactor Efficiency |
Contributor
There was a problem hiding this comment.
"Now that's a name I've not heard in a long time"
Just a nitpick, but I wonder if the term Reactor wouldn't be confusing both to LLMs and human users of this repo. It's been replaced long time ago. Search for "reactor" returns FSharpProjectOptionsReactor which is something entirely different.
Member
Author
There was a problem hiding this comment.
No you are right, I should get rid of it - its a risk of confusion (or lost tokens spend searching for what it is), while not gaining anything.
But it surprisingly came up when using 10y of PR feedback and issue comments in this repo as training data :))
Member
Author
There was a problem hiding this comment.
The training obviously overfitted on case of historical PRs receiving many comments, this is one of the examples.
(i.e. PR with 40 review comments is over-repsented in my training over one that was created without much negative feedback and merged it)
Will do another round.
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
2 participants
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.
No description provided.