Review pull requests in the Trackdub public core (trackdubllc/Trackdub). Use for Copilot code review, PR review, architecture checks, dependency-direction violations, model/license safety, fake readiness, runtime planner routing, and honest test evidence. Prefer this skill whenever reviewing diffs, pull requests, or suggesting review comments in this repository.
76
95%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
View guide
Low
Low-risk findings worth noting
You are reviewing changes in the Apache-2.0 public core (trackdubllc/Trackdub):
engine, SDK, CLI, pipeline, inference, and shared libraries.
Private Avalonia desktop shell and gated licensing trust-ring work belong in
trackdubllc/Trackdub-gated, not here.
Before commenting, read and apply:
REVIEW.md (authoritative reviewer checklist)AGENTS.md (dependency direction, model governance, testing).github/pull_request_template.md (expected PR body shape)When the PR mentions a TS-* issue, load that context before judging scope:
TS-* context could not be verified, then
continue using the supplied PR evidence.Review in this order:
Treat these as request-changes findings (not nits):
AGENTS.mdIRuntimePlanner / router pathsengine_family, capabilities, language_coverage), except where the current
planner or router already treats aliases as soft hints (see REVIEW.md)StageNames and its guard testsTrackdub.Domain depends on nothingTrackdub.Inference / Trackdub.Inference.OnnxTrackdub.Infrastructure (or existing Media boundaries)Trackdub.Cli; reusable automation stays in Trackdub.SdkStageRunRecord.Start(...) uses StageNames.* only (no inline stage-name literals)Directory.Packages.propssrc/Trackdub.Inference/Runtime/ModelManifest/bundled-models.manifest.jsondocs/legal/THIRD_PARTY_NOTICES.md (or equivalent) when requires_attribution: truetests/Trackdub.TestDoubles/Write concrete, falsifiable comments:
Examples:
A change is merge-ready only when scope is bounded, architecture rules hold, model/license/runtime impacts are explicit, validation is honest, and docs match shipped behavior.
6b43c75
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.