Repository navigation
feat(attention): add explainable attention score and make it the default sort - #59
Merged
Merged
Conversation
…ult sort Implements issue #33: - IAttentionScoreCalculator (new file, Needly.Application/Actions): AttentionScoreResult (0-100, clamped) + an ordered list of named AttentionScoreReason entries. Every point in the score is traceable to a reason; there is no opaque or hidden contribution. - AttentionScoreCalculator (new file, Needly.Infrastructure/Actions): a pure calculation over VisibleAction -- no direct DB or GitHub API access, so it is fast, cheap to call per-row, and easy to test. Deliberately scoped to signals VisibleAction already exposes today (action type, at-risk flag, waiting duration, direct vs team assignment, pinned, draft state, diff size bucket, labels). Review risk (#34) and agent authorship (#35) are explicitly NOT wired in yet -- those fields don't exist on VisibleAction until those issues land -- so this ships independently without touching NeedlyAction, ActionFilter, GitHubWebhookModels.cs or InboxVisibilityService.cs at all, avoiding any schema-level conflict with that concurrent work. - AttentionScoreOptions: every weight and threshold is configuration (IOptions-bound, validated at startup), not code, following the exact ActionRiskOptions/FollowUpOptions/DecideOptions convention already established in this codebase. - Inbox.razor.cs: the "Attention" sort order (previously just preserved InboxVisibilityService's fixed pinned/at-risk/waiting order) now actually ranks by computed score, falling back to oldest-waiting as the tie-break. This is still the default sort. - ActionRow: a score chip ("62 · High") with a hover tooltip listing every contributing reason, e.g. "+30 explicitly requested review · +20 at risk · +12 waiting 1 day". Known scope limits, by design: - TypeBasePoints (a Dictionary<string,int>) and WaitingTiers (a List<record>) are left at their C# defaults rather than also being wired into appsettings.json, to avoid unverified nested-collection configuration-binding risk; the scalar bonuses/penalties are bound and validated at startup. - Per-user weight customization (the issue's "configurable per user") is not built -- only per-installation defaults via configuration. A follow-up would need a settings UI and per-user storage. Fixes #33 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
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
1 participant
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.
Summary
Implements issue #33.
IAttentionScoreCalculator(new file,Needly.Application/Actions) —AttentionScoreResult(0-100, clamped) plus an ordered list of namedAttentionScoreReasonentries. Every point is traceable to a reason; nothing opaque.AttentionScoreCalculator(new file,Needly.Infrastructure/Actions) — a pure calculation overVisibleAction, no DB or GitHub API access, cheap to call per row. Scoped to signalsVisibleActionalready exposes today: action type, at-risk flag, waiting duration, direct-vs-team assignment, pinned, draft state, diff size bucket, labels.AttentionScoreOptions— every weight isIOptions-bound configuration, validated at startup, following the exactActionRiskOptions/FollowUpOptions/DecideOptionsconvention already in this codebase.Inbox.razor.cs— the "Attention" sort (previously just preservedInboxVisibilityService's fixed order) now actually ranks by computed score, oldest-waiting as tie-break. Still the default.ActionRow— a score chip (62 · High) with a hover tooltip listing every contributing reason, e.g.+30 explicitly requested review · +20 at risk · +12 waiting 1 day.Deliberately not wired in yet
Review risk (#34) and agent authorship (#35) are two of the signal families this score is meant to eventually include, but neither field exists on
VisibleActionyet — both issues are in progress concurrently. This PR does not touchNeedlyAction,ActionFilter,GitHubWebhookModels.csorInboxVisibilityService.csat all, so it has no schema-level conflict with that work. Once #34/#35 land, extending the calculator's input with their two extra fields is a small, low-risk follow-up.Known scope limits
TypeBasePoints(aDictionary<string,int>) andWaitingTiers(aList<record>) are left at their C# defaults rather than also being wired intoappsettings.json, to avoid unverified nested-collection configuration-binding risk. The scalar bonuses/penalties (at-risk, pinned, direct-assignment, draft, large-diff, urgent-label) are bound and validated at startup.Testing
dotnet build Needly.sln— 0 warnings, 0 errorsdotnet test Needly.sln— 388/388 passing (baseline 368 + 20 new)Fixes #33
🤖 Generated with Claude Code