What problem does it solve?
Maintainers facing a backlog of GitHub issues struggle to decide which reports belong together, which are duplicates, and who should diagnose each group. This Skill turns a raw issue set into evidence-backed, decision-coherent bundles with clear routing before any deep diagnosis begins.
Core Features & Use Cases
- Corpus retrieval and normalization: Batch-fetch compact issue metadata, treat public issue content as untrusted evidence, and normalize claims, version vectors, and report quality for each issue.
- Relationship and bundle analysis: Find evidence-backed relationships (duplicates, shared mechanisms, regression lineage) using stable signatures like errors, routes, and symbols, then group issues into independently diagnosable bundles.
- Routing and topology selection: Route each bundle to the main lane, native subagents, or independent Happier sessions, while keeping GitHub mutations separate and gated behind explicit authorization.
- Use Case: A maintainer asks to triage 30 open bug reports; the Skill fetches compact metadata, clusters them into four coherent bundles by likely mechanism, flags two duplicate candidates, and proposes a routing plan with version gaps and missing evidence.
Quick Start
Ask the assistant to triage a set of Happier GitHub issues by URL, label, or query and group them into diagnosis bundles with routing recommendations.
