Align vNext preview documentation truth

Co-authored-by: Sami Rusani <sr@samirusani>

Author: samrusani

.ai/handoff/CURRENT_STATE.md

@@ -52,18 +52,7 @@
 - `M-001` is implemented in this working tree and should be validated in GitHub Actions after PR push/merge.
 - Alice vNext Sprint 1 through Sprint 12, the live capture connectors sprint, and dogfood hardening are implemented and in the public-preview release gate.
 
-## Immediate Control Tower Decisions Needed
-- Use the separate `PostgresVNextStore` facade as the Sprint 2 persistence boundary unless a concrete integration blocker appears.
-- Decide whether old `memories` rows should be backfilled into full v2 canonical text/domain/sensitivity values before Sprint 2.
-- Decide whether the vNext capture API should expose local folder import directly or keep folder import CLI-only for local deployments.
-- Decide whether vNext context packs should be automatically persisted as generated artifacts or remain ephemeral until a user enqueues a queue task.
-- Decide which queue write policies may advance beyond deterministic local artifacts into external tool execution.
-- Decide how production daily/weekly scheduling should be configured and whether model-backed synthesis should replace or augment the deterministic Sprint 5 templates.
-- Decide how accepted vNext graph edges should influence retrieval ranking beyond trace/neighborhood visibility.
-- Decide how vNext belief status history should be unified with the existing temporal-state APIs and whether contradiction classification should use model-backed evidence review.
-- Decide how project-update rejection suppression should be persisted beyond event logs and how UI project pages should expose candidate review state.
-- Decide which remaining vNext UI write surfaces should become live-backed after the current live/fixture hybrid workspace.
-- Decide when the deterministic eval harness should graduate to model-backed, live-store-backed, or human-rated scoring.
-- Decide whether the next connector security layer should use OS keychain, hosted secret infrastructure, or both behind the existing secret-provider interface.
-- After dogfood hardening, decide the next build slice: broader live-backed `/vnext` workflows, managed connector OAuth, production scheduling, or model-backed evaluation.
+## Post-Preview Decisions
+- After the preview gate closes, choose the next product slice: broader live-backed `/vnext` workflows, managed connector OAuth, production scheduling, or model-backed evaluation.
+- Decide whether the older implementation-planning notes for early vNext sprints should be compacted or archived now that the release gate is the active execution posture.
 - Avoid reopening completed Phase 14 or `HF-001` scope unless a concrete defect or release-readiness issue is identified.

CURRENT_STATE.md

@@ -55,18 +55,7 @@ Canonical handoff state lives at [.ai/handoff/CURRENT_STATE.md](.ai/handoff/CURR
 - `M-001` is implemented in this working tree and should be validated in GitHub Actions after PR push/merge.
 - Alice vNext Sprint 1 through Sprint 12, the live capture connectors sprint, and dogfood hardening are implemented and in the public-preview release gate.
 
-## Immediate Control Tower Decisions Needed
-- Use the separate `PostgresVNextStore` facade as the Sprint 2 persistence boundary unless a concrete integration blocker appears.
-- Decide whether old `memories` rows should be backfilled into full v2 canonical text/domain/sensitivity values before Sprint 2.
-- Decide whether the vNext capture API should expose local folder import directly or keep folder import CLI-only for local deployments.
-- Decide whether vNext context packs should be automatically persisted as generated artifacts or remain ephemeral until a user enqueues a queue task.
-- Decide which queue write policies may advance beyond deterministic local artifacts into external tool execution.
-- Decide how production daily/weekly scheduling should be configured and whether model-backed synthesis should replace or augment the deterministic Sprint 5 templates.
-- Decide how accepted vNext graph edges should influence retrieval ranking beyond trace/neighborhood visibility.
-- Decide how vNext belief status history should be unified with the existing temporal-state APIs and whether contradiction classification should use model-backed evidence review.
-- Decide how project-update rejection suppression should be persisted beyond event logs and how UI project pages should expose candidate review state.
-- Decide which remaining vNext UI write surfaces should become live-backed after the current live/fixture hybrid workspace.
-- Decide when the deterministic eval harness should graduate to model-backed, live-store-backed, or human-rated scoring.
-- Decide whether the next connector security layer should use OS keychain, hosted secret infrastructure, or both behind the existing secret-provider interface.
-- After dogfood hardening, decide the next build slice: broader live-backed `/vnext` workflows, managed connector OAuth, production scheduling, or model-backed evaluation.
+## Post-Preview Decisions
+- After the preview gate closes, choose the next product slice: broader live-backed `/vnext` workflows, managed connector OAuth, production scheduling, or model-backed evaluation.
+- Decide whether the older implementation-planning notes for early vNext sprints should be compacted or archived now that the release gate is the active execution posture.
 - Avoid reopening completed Phase 14 or `HF-001` scope unless a concrete defect or release-readiness issue is identified.

README.md

@@ -41,11 +41,11 @@ Alice vNext is the next release candidate for the true second-brain product. It
 
 The vNext preview currently includes deterministic source capture, retrieval/context packs, queue/artifact workflows, daily and weekly brain artifacts, connection/contradiction/project/open-loop workflows, model-backed source-grounded synthesis, human artifact quality ratings, deterministic-vs-model comparison controls, synthetic evals, live local capture connectors for Telegram, local folders/Obsidian notes, browser clips, and Hermes/OpenClaw-style agent outputs, dedicated connector settings/state storage, encrypted local secret references, deterministic document connector payload ingestion, trusted agent memory commit with inline confirmation/review/reject policy, agent identity/policy auditing, a governed local scheduler with due scans, a local scheduler daemon, policy telemetry, dogfooding readiness telemetry, doctor/readiness checks, capture-to-brief traceability, and a live/fixture-backed `/vnext` operator workspace with source review, memory review, trusted commit audit, artifact review, project, open-loop, scheduler, connector, and doctor controls.
 
-## Public Alpha Quickstart
+## Public Preview Quickstart
 
 Alice is a local-first memory and continuity layer for humans and agents. It lets agents like Hermes, OpenClaw, or your own custom agents request scoped context, submit outputs, explicitly commit user-directed memories through Alice policy, propose reviewable memories, create open loops, and generate reviewable artifacts without giving them direct database access to trusted memory. The `/vnext` workspace is the operator console for review, audit, configuration, undo/correction/forget, and troubleshooting.
 
-Alice is not a notes app, an Obsidian clone, a chatbot with memory, hosted SaaS, or automatic memory autopilot. The public alpha is a technical local alpha for design partners and agent builders.
+Alice is not a notes app, an Obsidian clone, a chatbot with memory, hosted SaaS, or automatic memory autopilot. The current release candidate is a technical local public preview for design partners and agent builders.
 
 Fast path:
 
@@ -66,7 +66,7 @@ Then open:
 http://localhost:3000/vnext
 ```
 
-Load safe synthetic demo data and run the public alpha readiness gate:
+Load safe synthetic demo data and run the preview-readiness gate:
 
 ```bash
 alicebot vnext demo load --reset
@@ -78,16 +78,16 @@ alicebot vnext alpha check
 
 Agent integration starts with [docs/alpha/agent-integration.md](docs/alpha/agent-integration.md), [docs/alpha/mcp-tools.md](docs/alpha/mcp-tools.md), [docs/alpha/hermes-skill.md](docs/alpha/hermes-skill.md), and [docs/alpha/openclaw-skill.md](docs/alpha/openclaw-skill.md). Security, privacy, and limitations are documented in [docs/alpha/security-and-privacy.md](docs/alpha/security-and-privacy.md) and [docs/alpha/known-limitations.md](docs/alpha/known-limitations.md).
 
-Headless Ubuntu/Hermes dogfood starts with [docs/alpha/headless-ubuntu-install.md](docs/alpha/headless-ubuntu-install.md) and [docs/alpha/hermes-dogfood-ubuntu.md](docs/alpha/hermes-dogfood-ubuntu.md). The secure default is localhost binding plus SSH tunneling:
+Headless Ubuntu/Hermes preview dogfood starts with [docs/alpha/headless-ubuntu-install.md](docs/alpha/headless-ubuntu-install.md) and [docs/alpha/hermes-dogfood-ubuntu.md](docs/alpha/hermes-dogfood-ubuntu.md). The secure default is localhost binding plus SSH tunneling:
 
 ```bash
 ssh -L 3000:127.0.0.1:3000 -L 8000:127.0.0.1:8000 user@server
 ```
 
 Start with:
 
-- [Public alpha docs](docs/alpha/README.md)
-- [Public alpha quickstart](docs/alpha/quickstart.md)
+- [Public preview docs](docs/alpha/README.md)
+- [Public preview quickstart](docs/alpha/quickstart.md)
 - [First-run checklist](docs/alpha/first-run.md)
 - [Headless Ubuntu install](docs/alpha/headless-ubuntu-install.md)
 - [Hermes dogfood on Ubuntu](docs/alpha/hermes-dogfood-ubuntu.md)
@@ -108,7 +108,7 @@ Start with:
 - [Live capture connectors CTO summary](docs/vnext-live-capture-connectors-cto-summary.md)
 - [Dogfood hardening CTO summary](docs/vnext-dogfood-hardening-cto-summary.md)
 - [Live-backed operator console CTO summary](docs/vnext-live-backed-operator-console-cto-summary.md)
-- [Public alpha packaging CTO summary](docs/vnext-public-alpha-packaging-cto-summary.md)
+- [Legacy public-alpha packaging CTO summary](docs/vnext-public-alpha-packaging-cto-summary.md)
 - [Headless Ubuntu packaging CTO summary](docs/vnext-headless-ubuntu-cto-summary.md)
 - [Agentic memory commit CTO summary](docs/vnext-agentic-memory-commit-cto-summary.md)
 - [Dogfood daily checklist](docs/runbooks/vnext-dogfood-daily-checklist.md)

ROADMAP.md

@@ -5,7 +5,6 @@
 - Phase 10: shipped
 - Phase 11: shipped
 - Bridge `B1`-`B4`: shipped
-- Bridge Phase (`B1`-`B4`): shipped
 - Phase 12: shipped
 - Phase 13: shipped
 - Phase 14: shipped

RULES.md

@@ -8,8 +8,11 @@
 - Keep [CURRENT_STATE.md](CURRENT_STATE.md) factual/current and [ROADMAP.md](ROADMAP.md) future-facing.
 - Keep [.ai/handoff/CURRENT_STATE.md](.ai/handoff/CURRENT_STATE.md) as the canonical handoff copy when duplicate current-state files exist.
 
-## Phase 14 Rule
-- Phase 14 is a platform-and-adoption phase. Prioritize provider adapters, model packs, reference integrations, and design partner onboarding. Do not allow scope drift into new substrate research, new channels, or enterprise governance work unless required by a declared Phase 14 deliverable.
+## Phase 14 Baseline Rule
+- Phase 14 was a platform-and-adoption phase. Preserve that shipped boundary as baseline truth: provider adapters, model packs, reference integrations, and design partner onboarding are complete baseline surfaces, not open scope.
+
+## Active vNext Preview Rule
+- While `v0.5.1-vnext-preview` is the active release gate, prioritize release accuracy, install clarity, preview-safe docs, and verification evidence. Do not add net-new feature scope unless it is required to fix preview readiness, release evidence, or operator safety.
 
 ## Provider Rules
 - Continuity semantics must not fork by provider.

docs/alpha/README.md

@@ -1,6 +1,8 @@
-# Alice vNext Public Alpha
+# Alice vNext Public Preview
 
-Alice vNext public alpha is a technical, local-first package for design partners who want agent memory and continuity without hosted storage or direct database writes by agents.
+This folder keeps the legacy `docs/alpha` path name, but the current release posture is Alice vNext public preview.
+
+Alice vNext public preview is a technical, local-first package for design partners who want agent memory and continuity without hosted storage or direct database writes by agents.
 
 Alice is agent-first, not dashboard-first:
 

docs/alpha/agent-integration.md

@@ -38,7 +38,7 @@ Permission profiles:
 ## CLI Example
 
 ```bash
-alicebot context-pack "Alice public alpha sprint context" --domain project --project Alice
+alicebot context-pack "Alice public preview sprint context" --domain project --project Alice
 
 alicebot vnext agents ingest-output \
   --agent-id openclaw \
@@ -51,7 +51,7 @@ alicebot vnext agents ingest-output \
   --domain project \
   --sensitivity private \
   --propose-memory \
-  "Decision: Alice public alpha agents use scoped context packs and review-only memory proposals."
+  "Decision: Alice public preview agents use scoped context packs and review-only memory proposals."
 
 alicebot vnext memories commit \
   --agent-id openclaw \
@@ -60,7 +60,7 @@ alicebot vnext memories commit \
   --project-scope Alice \
   --permission-profile project_scoped_agent \
   --title "Release gate decision" \
-  --text "Alice public alpha release gates require doctor, smokes, evals, and git diff checks before merge." \
+  --text "Alice public preview release gates require doctor, smokes, evals, and git diff checks before merge." \
   --domain project \
   --sensitivity private \
   --confidence 0.94

docs/alpha/demo-mode.md

@@ -1,6 +1,6 @@
 # Demo Dataset And Demo Mode
 
-The public alpha ships a small synthetic vNext dataset in:
+The public preview ships a small synthetic vNext dataset in:
 
 ```text
 fixtures/vnext/demo_dataset.json

docs/alpha/dogfooding-guide.md

@@ -1,6 +1,6 @@
 # Dogfooding Guide
 
-Daily public alpha dogfood loop:
+Daily public preview dogfood loop:
 
 ```bash
 alicebot vnext doctor --fix-safe --ci
@@ -21,7 +21,7 @@ Then use `/vnext`:
 Include these with bug reports:
 
 - output from `alicebot vnext doctor --fix-safe --ci`
-- output from `alicebot vnext alpha check --skip-smokes`
+- output from the preview-readiness command `alicebot vnext alpha check --skip-smokes`
 - failing smoke command and error
 - `/vnext` page and action that failed
 - redacted connector config, never secret values

docs/alpha/first-run.md

@@ -10,7 +10,7 @@ Use this checklist after a fresh clone.
 | 4. Start API | `APP_RELOAD=false ./scripts/api_dev.sh` | API serves locally |
 | 5. Start scheduler | `alicebot vnext scheduler daemon start --foreground` | daemon reports status and due scans |
 | 6. Check live browser env | `cat apps/web/.env.local` | `NEXT_PUBLIC_ALICEBOT_API_BASE_URL=http://127.0.0.1:8000` and a local user id are present |
-| 7. Start `/vnext` | `pnpm --dir apps/web dev` | web app serves locally |
+| 7. Start `/vnext` | `pnpm --dir apps/web build` then `pnpm --dir apps/web start --hostname 127.0.0.1 --port 3000` | web app serves locally |
 | 8. Open `/vnext` | `http://localhost:3000/vnext` | live mode loads or shows a clear empty state |
 | 9. Configure Brain Charter | [../vnext/ALICE.example.md](../vnext/ALICE.example.md) | Brain Charter is visible in Settings |
 | 10. Configure one capture path | local folder, browser clipper, or Telegram | connector health becomes configured |
@@ -25,3 +25,5 @@ Use this checklist after a fresh clone.
 When doctor fails, go to [doctor.md](doctor.md) first. It lists the exact command that usually fixes the blocking issue.
 
 For local live mode, `.env` must include `CORS_ALLOWED_ORIGINS=http://127.0.0.1:3000,http://localhost:3000`, and `apps/web/.env.local` must include `NEXT_PUBLIC_ALICEBOT_API_BASE_URL=http://127.0.0.1:8000`.
+
+Use `pnpm --dir apps/web dev` only while editing the web UI. For long-running agent or Hermes sessions, prefer `make runtime` as the combined API plus web command, or use the `next start` command above when API is already running.