epic: beta.3 → stable roadmap re-forecast

[rickylabs]

Summary

Re-forecast the whole 0.0.1 roadmap Aspire-style: each remaining release cut (beta.3, beta.4, beta.5, 0.0.1-stable) is designed from epics, requirements, and deep evidence analysis — not by folding whatever happens to be closeable into the next beta.

Trigger: label-trust proved unsafe — #260 was closed with its gate:e2e checkbox unchecked and the AI-plugin e2e was never wired (now #388). This program applies evidence over labels: every claim of shipped scope is verified in code before it counts toward a milestone.

Part of #301 (Road to 0.0.1-stable). Dogfoods harness V3 (#389): tracked run dir, draft-PR-on-start, status: stage lifecycle, tiered agent model, close-gate #387.

Pillars

1. Evidence sweep — every open issue (real remaining scope, dedup, mis-milestoned, stale, already-done-but-open); the 21 PRs merged into beta.2 (delivered vs claimed, false-closed acceptance); true completion state of the open epics (epic: Road to 0.0.1-stable #301, epic: NetScript AI Stack — first-class AI runtime, chat & plugin seams (anchor #219) #238 + [AI-stack] plugins/ai: state-of-the-art flagship parity — e2e, verify-plugin, /v1/ai impl+tests, doctor/scaffolder coverage #388, epic: NetScript enterprise deployment framework (cloud-agnostic + bare-metal, CLI + Aspire) #327, epic: Agentic Workflow Doctrine V3 (harness V3) #389, Prisma-Next DB migration, docs overhaul).
2. Code + architecture verification — verify claimed-shipped work exists and is sound (AI e2e wiring, /v1/ai surface, deploy targets); identify genuine blockers to a cuttable, dogfoodable 0.0.1-stable that eis-chat can build on.
3. Milestone re-forecast — themed scope + realistic ETA for beta.3 / beta.4 / beta.5 / 0.0.1-stable; reconcile the moves already applied ([AI-stack] plugins/ai: state-of-the-art flagship parity — e2e, verify-plugin, /v1/ai impl+tests, doctor/scaffolder coverage #388/process: gate issue closure on verified acceptance (stop false-closed acceptance) #387→beta.3, [AI-stack P5] plugin-ai: opt-in --gateway centralized AI service #262/[AI-stack P2-follow] plugins/ai --mcp / skill scaffolder + e2e variant (needs E5 SkillLoaderPort) #290/[AI-stack E8] @netscript/ai: orchestration primitives (fan-out + bounded-cycle) #247→beta.4, [AI-stack anchor] streams: no durable-CHAT integration — @netscript/fresh/ai seam (StreamDB shapes → Electric "Decoding failed") #219→beta.3, epic: Agentic Workflow Doctrine V3 (harness V3) #389→stable); apply reversible milestone/label re-assignments; surface the strategic roadmap for owner ratification.

Checklist

• S1 Evidence sweep: open issues
• S2 Evidence sweep: beta.2 merged PRs (delivered vs claimed)
• S3 Evidence sweep: epic completion states
• S4 Code verification dive (AI e2e, /v1/ai, deploy targets, flagship-AI bar)
• S5 Roadmap artifact: per-milestone theme + scope + ETA + rationale + evidence trail
• S6 GitHub reconciliation: milestones/labels re-assigned to match the forecast
• S7 Owner ratification of the strategic roadmap (stable definition, cut sequence)

Links

• Parent: epic: Road to 0.0.1-stable #301 (Road to 0.0.1-stable)
• Harness V3 program: epic: Agentic Workflow Doctrine V3 (harness V3) #389
• Close-gate mechanism: process: gate issue closure on verified acceptance (stop false-closed acceptance) #387 · Exemplar false-close: [AI-stack P2] @netscript/plugin-ai manifest + connector + scaffold + in-process emitters #260 → [AI-stack] plugins/ai: state-of-the-art flagship parity — e2e, verify-plugin, /v1/ai impl+tests, doctor/scaffolder coverage #388

Non-goals

• No framework/plugin source edits — code defects found by the sweep are filed as issues, not fixed here.
• Strategic decisions (stable definition, cut order) are proposed, not finalized, pending owner sign-off.

Do not put closing keywords on this epic; it closes by hand when all children are done.

[rickylabs]

Bootstrap complete: draft PR #392 (status:research), tracked run dir .llm/runs/chore-roadmap-beta3-stable-reforecast--reforecast/ on branch chore/roadmap-beta3-stable-reforecast. G1 evidence sweeps (Opus 4.8 x3: open issues / beta.2 PR delivery / epic+code verification) in flight.

[rickylabs]

Roadmap re-forecast — status (2026-07-04)

Evidence sweep + forecast complete; GitHub reconciled; PLAN-EVAL dispatched to OpenHands.

• Draft PR chore(roadmap): beta.3->stable roadmap re-forecast (V3 dogfood) #392 opened; run dir .llm/runs/chore-roadmap-beta3-stable-reforecast--reforecast/
• G1 evidence: 3 Opus 4.8 sweeps (open issues / beta.2 delivery / epic+code)
• G2 synthesis: roadmap-0.0.1.md (4 themed cuts + ETAs + owner questions)
• G3 apply: beta.5 milestone created; ~20 reversible milestone moves; bugs bug(deploy): Aspire compose target not registered in DEFAULT_DEPLOY_TARGETS (docker/compose lane dead) #393/test(deploy): no deploy target has any e2e coverage (add deploy e2e gate) #394 filed; taxonomy fixes feat(aspire): first-party deno desktop app support in the generator (native window as a dev-stack resource) #375/bug(plugin-workers): built-in workers-plugin-health-check job entrypoint not resolvable under the thin-plugin surface (Module not found) #376
• G4 PLAN-EVAL (OpenHands, dispatched — awaiting verdict)
• G5 owner ratification (5 strategic questions in roadmap §5)

Forecast: beta.3 AI-plugin parity + acceptance integrity (#388/#387/#380/#379/#219/#376/#393/#394) · beta.4 AI depth · beta.5 deploy + maturity (incl. #389) · stable leadership bench + prod gates. Single load-bearing dogfood gap = #388.

Part of #391.