Problem
Maintainers care about review load, stale queues, duplicate clusters, and reviewable PRs. Individual PR advisories are not enough to understand repo-level burden.
Desired behavior
Generate repo-level burden forecast snapshots and expose them through repo intelligence and private MCP/API surfaces.
Implementation notes
- Use open PR age, stale PRs, duplicate clusters, reviewable PRs, failing checks, broad diffs, and unlinked PRs.
- Generate snapshots asynchronously; do not compute large repo forecasts inline.
- Keep
ragflow/sure scale in mind.
- Store enough freshness metadata for clients to explain when forecasts are stale.
Public/private boundaries
Forecasts are private API/MCP intelligence. Public GitHub comments should stay minimal and sanitized.
Acceptance criteria
- Large queue fixtures complete within Worker-safe bounds.
- Forecasts identify top burden sources and cleanup actions.
- Repo intelligence includes freshness and data-quality warnings.
- No broad request-time scans are needed.
Tests
- Small repo queue fixture.
- Large
ragflow/sure-style queue fixture.
- Duplicate cluster fixture.
- Stale PR trend fixture.
- Snapshot freshness fixture.
- Performance regression for bounded work.
Problem
Maintainers care about review load, stale queues, duplicate clusters, and reviewable PRs. Individual PR advisories are not enough to understand repo-level burden.
Desired behavior
Generate repo-level burden forecast snapshots and expose them through repo intelligence and private MCP/API surfaces.
Implementation notes
ragflow/surescale in mind.Public/private boundaries
Forecasts are private API/MCP intelligence. Public GitHub comments should stay minimal and sanitized.
Acceptance criteria
Tests
ragflow/sure-style queue fixture.