aggregator: derive epic ETA bands and explainable stall risk

Parent chain: #1184 → #1256 → this issue. GitHub currently caps #1184 at 100 direct sub-issues and rejects additional links (REST 422), so implementation slices are nested under #1256, already a native child of #1184; no existing parent links were changed.

Design context: #1256; doc/2026-11-01T21-07-00Z_design_pm-stack_forecasting/2026-11-01T21-07-00Z_design_pm-stack_forecasting.md (§1, §3). Depends on the reusable item-history source in #3981. Coordinate with #1281: reuse its history/throughput seam, but do not duplicate its charts.

Goal

Add pure, deterministic derivations for project/epic ETA distributions and item stall-risk explanations. The current project/coverage inputs do not contain enough item-level history to calculate these values.

Acceptance criteria

ETA uses complete observed weekly item-history buckets, including zero-completion weeks, over trailing 4-week and 8-week windows; incomplete windows are excluded and the scope/sample are returned.
One direct native child closure is the documented default unit; merged PRs do not double-count. Reopened issues are not counted as currently complete. Existing effort weights and open-ended/unsized work remain explicit and never create invented precision.
With fewer than 3 completions or 4 complete weeks, the result is low-confidence/borrowed from a clearly named comparable scope or “not enough history”; with no defensible baseline, no ETA date is returned. Numeric forecasts include P50/P80 ranges, confidence, input coverage, and model version.
Stall risk is a versioned, explainable 0–100 rule score with contributing reason codes/timestamps; first-observed blocker age is labelled a lower bound, and unavailable heartbeat/requeue signals stay unknown.
Score ≥60 is surfaced as informational “at risk”; score alone does not create a Needs-you escalation. Tests cover missing history, zero pace, reopen, sparse samples, borrowed baseline, and each risk signal.

agent-note:v1
state: triaged (coverage pass)
Coverage pass 2026-11-03: reviewed in the fleet-wide per-issue pass (drift / overlap / already-handled / stale / blockers / ordering). Verdict: still live; queued in repo-health-code-pilot as triage at priority 40 (priority/normal). If this is already handled at current main, the triage closes it with evidence; if it needs splitting, the triage files the split.
Found via: coverage pass 2026-11-03

🤖 Agent note · picked up

Agent: queue worker (repo-worker-code), lane code-e
Session / ID: task 8962
Found via: queue claim
Task: task 8962 for issue aggregator: derive epic ETA bands and explainable stall risk #3982 (small_fix) — execution lane assigned direct; this lane works it per the runbook
Expectation: work proceeds in checkpoints to a verified terminal state (complete / park / block)
Links: issue aggregator: derive epic ETA bands and explainable stall risk #3982

aggregator(forecast): compare as-of timestamps as instants, not strin…

🤖 Agent note · parked — CI pending

Agent: queue worker (repo-worker-code), lane code-e
Session / ID: task 8962
Found via: queue claim (resume of parked task; PR aggregator(forecast): derive epic ETA bands and explainable stall risk (#3982) #5236 @40ed243f2, stage awaiting_ci)
Task: all 5 Codex findings on PR aggregator(forecast): derive epic ETA bands and explainable stall risk (#3982) #5236 dispositioned — findings 1–4 were already fixed in 40ed243 with tests (replies posted on each thread); round-2 finding (forecast.ts:155, as-of string compare) fixed in e10e253 with 3 new regression tests. bun test forecast-3982 on the laptop: 23 pass, 0 fail. New head e10e253 pushed.
Park: CI on PR aggregator(forecast): derive epic ETA bands and explainable stall risk (#3982) #5236 (contract corpus; aggregator tests and typecheck) still QUEUED after a 5-minute poll — task 8962 parked 20 minutes. Resume: if checks green, merge PR aggregator(forecast): derive epic ETA bands and explainable stall risk (#3982) #5236 and close issue aggregator: derive epic ETA bands and explainable stall risk #3982; if red, fix on branch queue/8962-3982-epic-eta.
Links: issue aggregator: derive epic ETA bands and explainable stall risk #3982, PR aggregator(forecast): derive epic ETA bands and explainable stall risk (#3982) #5236

aggregator(forecast): bound blocker age to effective asOf; unknown re…

Merge pull request #5236 from developing-today/queue/8962-3982-epic-eta

🤖 Agent note · done

Agent: queue worker (repo-worker-code), lane code-a
Session / ID: task 8962
Result: fixed_merged — PR aggregator(forecast): derive epic ETA bands and explainable stall risk (#3982) #5236 merged as eb36246. Closing issue aggregator: derive epic ETA bands and explainable stall risk #3982 on the merge evidence (the PR did not auto-close it).
Resume work this shift: revalidated PR aggregator(forecast): derive epic ETA bands and explainable stall risk (#3982) #5236 at head e10e253; found two undispositioned Codex round-3 P2 findings and fixed them in ee57a97 (blocker-age history bounded to the effective asOf; explicitly requested children absent from the latest complete cutoff are now reported as state-unknown scope limitations and block the nothing-left completion branch), with regression tests and replies on both threads. Codex round-4 P2 (resampling estimator per design §1) dispositioned on its thread as a model change and filed as follow-up issue aggregator(forecast): resampling estimator for epic ETA (design §1) — zero-median bursty buckets return no_measurable_pace #5274.
Tests: bun test test/forecast-3982.test.ts on the laptop — 25 pass, 0 fail. Full aggregator suite: 1486 pass; the only failures are alerts/notifications suites failing to resolve the web-push package because this worktree has no node_modules install (empty directory) — pre-existing/environmental, disjoint from this diff (forecast.ts + its test only).
Links: task 8962 · issue aggregator: derive epic ETA bands and explainable stall risk #3982 · PR aggregator(forecast): derive epic ETA bands and explainable stall risk (#3982) #5236 · follow-up issue aggregator(forecast): resampling estimator for epic ETA (design §1) — zero-median bursty buckets return no_measurable_pace #5274

Navigation Menu

Repository navigation

aggregator: derive epic ETA bands and explainable stall risk #3982

Goal

Acceptance criteria

65 remaining items

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

aggregator: derive epic ETA bands and explainable stall risk #3982

Description

Goal

Acceptance criteria

Activity

developing-today commented on Nov 3, 2026

65 remaining items

dezren39 commented on Nov 4, 2026

dezren39 commented on Nov 4, 2026

dezren39 commented on Nov 4, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions