498 lines
27 KiB
Markdown
498 lines
27 KiB
Markdown
# LetsBe Biz — Implementation Plan
|
||
|
||
**Date:** February 27, 2026
|
||
**Team:** Claude Opus 4.6 Architecture Team
|
||
**Document:** 04 of 09
|
||
**Status:** Proposal — Competing with independent team
|
||
|
||
---
|
||
|
||
## Table of Contents
|
||
|
||
1. [Phase Overview](#1-phase-overview)
|
||
2. [Phase 1 — Foundation (Weeks 1-4)](#2-phase-1--foundation-weeks-1-4)
|
||
3. [Phase 2 — Integration (Weeks 5-8)](#3-phase-2--integration-weeks-5-8)
|
||
4. [Phase 3 — Customer Experience (Weeks 9-12)](#4-phase-3--customer-experience-weeks-9-12)
|
||
5. [Phase 4 — Polish & Launch (Weeks 13-16)](#5-phase-4--polish--launch-weeks-13-16)
|
||
6. [Dependency Graph](#6-dependency-graph)
|
||
7. [Parallel Workstreams](#7-parallel-workstreams)
|
||
8. [Scope Cut Table](#8-scope-cut-table)
|
||
9. [Critical Path](#9-critical-path)
|
||
|
||
---
|
||
|
||
## 1. Phase Overview
|
||
|
||
```
|
||
Week 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
|
||
├────────────────┤
|
||
│ PHASE 1: │
|
||
│ Foundation │
|
||
│ Safety Wrapper │
|
||
│ Secrets Proxy │
|
||
│ P0 Tests │
|
||
│ ├────────────────┤
|
||
│ │ PHASE 2: │
|
||
│ │ Integration │
|
||
│ │ Hub APIs │
|
||
│ │ Tool Adapters │
|
||
│ │ Browser Tool │
|
||
│ │ ├────────────────┤
|
||
│ │ │ PHASE 3: │
|
||
│ │ │ Customer UX │
|
||
│ │ │ Mobile App │
|
||
│ │ │ Provisioner │
|
||
│ │ │ ├────────────────┤
|
||
│ │ │ │ PHASE 4: │
|
||
│ │ │ │ Polish │
|
||
│ │ │ │ Security Audit│
|
||
│ │ │ │ Launch │
|
||
```
|
||
|
||
| Phase | Duration | Focus | Exit Criteria |
|
||
|-------|----------|-------|---------------|
|
||
| 1 | Weeks 1-4 | Safety Wrapper + Secrets Proxy core | Secrets redaction passes all P0 tests; command classification works; OpenClaw routes through wrapper |
|
||
| 2 | Weeks 5-8 | Hub APIs + tool adapters + billing | Hub ↔ Safety Wrapper protocol working; 6 P0 tool adapters operational; token metering flowing to billing |
|
||
| 3 | Weeks 9-12 | Mobile app + customer portal + provisioner | End-to-end: payment → provision → AI ready → mobile chat working |
|
||
| 4 | Weeks 13-16 | Security audit + polish + launch | Founding member launch: first 10 customers onboarded |
|
||
|
||
---
|
||
|
||
## 2. Phase 1 — Foundation (Weeks 1-4)
|
||
|
||
### Goal: Safety Wrapper and Secrets Proxy functional with comprehensive P0 tests
|
||
|
||
#### Week 1: Safety Wrapper Skeleton + Secrets Registry
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 1.1 Monorepo setup (Turborepo, packages structure) | 2d | Working monorepo with packages/safety-wrapper, packages/secrets-proxy, packages/shared-types | — |
|
||
| 1.2 Safety Wrapper HTTP server skeleton | 2d | Express/Fastify server on localhost:8200 with health endpoint | 1.1 |
|
||
| 1.3 SQLite schema + migration system | 1d | secrets, approvals, audit_log, token_usage, hub_state tables | 1.1 |
|
||
| 1.4 Secrets registry implementation | 3d | ChaCha20-Poly1305 encrypted SQLite vault; CRUD operations; pattern generation | 1.3 |
|
||
| 1.5 Tool execution endpoint (POST /api/v1/tools/execute) | 2d | Request parsing, validation, routing to executors | 1.2 |
|
||
|
||
#### Week 2: Command Classification + Tool Executors
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 2.1 Command classification engine | 3d | Deterministic rule engine for all 5 tiers; shell command classifier with allowlist | 1.5 |
|
||
| 2.2 Shell executor (port from sysadmin agent) | 2d | execFile-based execution with path validation, timeout, metacharacter blocking | 2.1 |
|
||
| 2.3 Docker executor | 1d | Docker subcommand classifier + executor | 2.2 |
|
||
| 2.4 File read/write executor | 1d | Path traversal prevention, size limits, atomic writes | 2.2 |
|
||
| 2.5 Env read/update executor | 1d | .env parsing, atomic update with temp→rename | 2.2 |
|
||
| 2.6 P0 tests: command classification | 2d | 100+ test cases covering all tiers, edge cases, shell metacharacters | 2.1 |
|
||
|
||
#### Week 3: Secrets Proxy + Redaction Pipeline
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 3.1 Secrets Proxy HTTP server | 1d | Transparent proxy on localhost:8100 | 1.1 |
|
||
| 3.2 Layer 1: Aho-Corasick registry redaction | 2d | O(n) multi-pattern matching against all known secrets | 1.4, 3.1 |
|
||
| 3.3 Layer 2: Regex safety net | 1d | Private keys, JWTs, bcrypt, connection strings, env patterns | 3.1 |
|
||
| 3.4 Layer 3: Shannon entropy filter | 1d | High-entropy blob detection (≥4.5 bits, ≥32 chars) | 3.1 |
|
||
| 3.5 Layer 4: JSON key scanning | 0.5d | Sensitive key name detection in JSON payloads | 3.1 |
|
||
| 3.6 P0 tests: secrets redaction | 2.5d | TDD — test matrix from Technical Architecture §19.2: registry match, patterns, entropy, false positives, performance (<10ms) | 3.2-3.5 |
|
||
|
||
#### Week 4: Autonomy Engine + OpenClaw Integration
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 4.1 Autonomy resolution engine | 2d | Level 1/2/3 gating matrix; per-agent overrides; external comms gate | 2.1 |
|
||
| 4.2 Approval queue (local) | 1d | SQLite-backed pending approvals with expiry | 4.1 |
|
||
| 4.3 Credential injection (SECRET_REF resolution) | 2d | Intercept SECRET_REF placeholders, inject real values from registry | 1.4, 2.2 |
|
||
| 4.4 OpenClaw integration: configure tool routing | 2d | OpenClaw routes tool calls to Safety Wrapper HTTP API | 4.3 |
|
||
| 4.5 OpenClaw integration: configure LLM proxy | 1d | OpenClaw routes LLM calls through Secrets Proxy (port 8100) | 3.1 |
|
||
| 4.6 P0 tests: autonomy level mapping | 1d | All 3 levels × 5 tiers × per-agent override scenarios | 4.1 |
|
||
| 4.7 Integration test: OpenClaw → Safety Wrapper → tool execution | 1d | End-to-end tool call with classification, gating, execution, audit logging | 4.4 |
|
||
|
||
### Phase 1 Exit Criteria
|
||
|
||
- [ ] Secrets Proxy redacts all known secret patterns with <10ms latency
|
||
- [ ] Command classifier correctly tiers all defined tools + shell commands
|
||
- [ ] Autonomy engine correctly gates/executes at all 3 levels
|
||
- [ ] OpenClaw successfully routes tool calls through Safety Wrapper
|
||
- [ ] OpenClaw successfully routes LLM calls through Secrets Proxy
|
||
- [ ] SECRET_REF injection works for tool execution
|
||
- [ ] All P0 tests pass (secrets redaction, command classification, autonomy mapping)
|
||
- [ ] Audit log records every tool call
|
||
|
||
---
|
||
|
||
## 3. Phase 2 — Integration (Weeks 5-8)
|
||
|
||
### Goal: Hub ↔ Safety Wrapper protocol, P0 tool adapters, billing pipeline
|
||
|
||
#### Week 5: Hub Communication Protocol
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 5.1 Hub: /api/v1/tenant/register endpoint | 1d | Registration token validation, API key generation | Phase 1 |
|
||
| 5.2 Hub: /api/v1/tenant/heartbeat endpoint | 2d | Metrics ingestion, config response, pending commands | 5.1 |
|
||
| 5.3 Hub: /api/v1/tenant/config endpoint | 1d | Full config delivery (agents, autonomy, classification) | 5.1 |
|
||
| 5.4 Safety Wrapper: Hub client implementation | 2d | Registration, heartbeat loop, config sync, backoff/jitter | 5.1-5.3 |
|
||
| 5.5 Hub: ServerConnection model update | 0.5d | Add safetyWrapperUrl, openclawVersion, configVersion fields | — |
|
||
| 5.6 P1 tests: Hub ↔ Safety Wrapper protocol | 1.5d | Registration, heartbeat, config sync, network failure handling | 5.4 |
|
||
|
||
#### Week 6: Token Metering + Billing
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 6.1 Safety Wrapper: token metering capture | 2d | Capture from OpenRouter response headers; hourly bucket aggregation | Phase 1 |
|
||
| 6.2 Hub: TokenUsageBucket + BillingPeriod models | 1d | Prisma migration, model definitions | — |
|
||
| 6.3 Hub: /api/v1/tenant/usage endpoint | 1d | Ingest usage buckets, update billing period | 6.2 |
|
||
| 6.4 Hub: /api/v1/admin/billing/* endpoints | 2d | Customer billing summary, history, overage trigger | 6.2 |
|
||
| 6.5 Stripe Billing Meters integration | 2d | Overage metering + premium model metering via Stripe | 6.4 |
|
||
| 6.6 Hub: FoundingMember model + multiplier logic | 1d | Token multiplier applied to billing period creation | 6.2 |
|
||
| 6.7 Hub: usage alerts (80/90/100%) | 1d | Trigger push notifications at pool thresholds | 6.3 |
|
||
|
||
#### Week 7: Tool Adapters (P0)
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 7.1 Tool registry template + generator | 1d | tool-registry.json generation from provisioner env files | Phase 1 |
|
||
| 7.2 Master skill (SKILL.md) | 0.5d | Teach AI three access patterns (API, CLI, browser) | 7.1 |
|
||
| 7.3 Cheat sheet: Portainer | 0.5d | REST v2 API endpoints for container management | — |
|
||
| 7.4 Cheat sheet: Nextcloud | 1d | WebDAV + OCS REST endpoints | — |
|
||
| 7.5 Cheat sheet: Chatwoot | 1d | REST v1/v2 endpoints for conversation management | — |
|
||
| 7.6 Cheat sheet: Ghost | 0.5d | Content + Admin REST endpoints | — |
|
||
| 7.7 Cheat sheet: Cal.com | 0.5d | REST v2 endpoints | — |
|
||
| 7.8 Cheat sheet: Stalwart Mail | 0.5d | REST endpoints for account/domain management | — |
|
||
| 7.9 Integration tests: agent → tool via Safety Wrapper | 2d | 6 tools: API call with SECRET_REF, classification, execution, response | 7.3-7.8 |
|
||
|
||
#### Week 8: Approval Queue + Config Sync
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 8.1 Hub: CommandApproval model + endpoints | 2d | CRUD for approvals; customer + admin approval endpoints | 6.2 |
|
||
| 8.2 Hub: /api/v1/tenant/approval-request endpoint | 1d | Safety Wrapper pushes approval requests to Hub | 8.1 |
|
||
| 8.3 Hub: /api/v1/tenant/approval-response/{id} endpoint | 1d | Safety Wrapper polls for approval decisions | 8.1 |
|
||
| 8.4 Hub: AgentConfig model + admin endpoints | 2d | CRUD for agent configs; sync to Safety Wrapper | — |
|
||
| 8.5 Config sync: Hub → Safety Wrapper | 1d | Config versioning; delta delivery via heartbeat | 5.2, 8.4 |
|
||
| 8.6 Push notification service skeleton | 1d | Expo Push token registration; notification sending | — |
|
||
| 8.7 Integration test: approval round-trip | 1d | Red command → gate → push to Hub → approve → execute | 8.3 |
|
||
|
||
### Phase 2 Exit Criteria
|
||
|
||
- [ ] Safety Wrapper registers with Hub and maintains heartbeat
|
||
- [ ] Token usage flows from Safety Wrapper → Hub → BillingPeriod
|
||
- [ ] Stripe overage billing triggers when pool exhausted
|
||
- [ ] 6 P0 tool cheat sheets operational (agent can use Portainer, Nextcloud, Chatwoot, Ghost, Cal.com, Stalwart)
|
||
- [ ] Approval round-trip works: gate → Hub → approve → execute
|
||
- [ ] Config sync: Hub agent config changes propagate to Safety Wrapper
|
||
- [ ] Founding member multiplier applies to billing periods
|
||
|
||
---
|
||
|
||
## 4. Phase 3 — Customer Experience (Weeks 9-12)
|
||
|
||
### Goal: End-to-end customer journey from payment to mobile chat
|
||
|
||
#### Week 9: Mobile App Foundation
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 9.1 Expo project setup (Bare Workflow, SDK 52) | 1d | Project scaffolding, EAS configuration | — |
|
||
| 9.2 Auth flow (login, JWT storage) | 2d | Login screen, secure token storage, auto-refresh | — |
|
||
| 9.3 Chat view with SSE streaming | 3d | Real-time agent response rendering via Hub relay | Phase 2 |
|
||
| 9.4 Agent selector (team chat vs. direct) | 1d | Agent roster, tap to open direct chat | 9.3 |
|
||
| 9.5 Push notification setup (Expo Push) | 1d | Token registration, notification categories, background handlers | — |
|
||
| 9.6 Approval cards with one-tap approve/deny | 1d | In-app queue + push notification action buttons | 9.5, Phase 2 |
|
||
|
||
#### Week 10: Customer Portal + Chat Relay
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 10.1 Hub: customer portal API (/api/v1/customer/*) | 3d | Dashboard, agents, usage, approvals, tools, billing endpoints | Phase 2 |
|
||
| 10.2 Hub: chat relay service | 2d | App → Hub → Safety Wrapper → OpenClaw → response stream | Phase 2 |
|
||
| 10.3 Hub: WebSocket endpoint for real-time chat | 2d | Persistent connection for chat + notification delivery | 10.2 |
|
||
| 10.4 Mobile: dashboard screen | 1d | Server status, morning briefing, quick actions | 10.1 |
|
||
| 10.5 Mobile: usage dashboard | 1d | Per-agent, per-model token usage with trends | 10.1 |
|
||
|
||
#### Week 11: Provisioner Update + Website
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 11.1 Provisioner: update step 10 for OpenClaw + Safety Wrapper | 3d | Deploy LetsBe AI stack, generate configs, seed secrets | Phase 1 |
|
||
| 11.2 Provisioner: n8n cleanup | 1d | Remove all n8n references (7 files) | — |
|
||
| 11.3 Provisioner: config.json cleanup (CRITICAL fix) | 0.5d | Remove plaintext passwords post-provisioning | — |
|
||
| 11.4 Website: landing page + onboarding flow pages 1-5 | 2d | Business description → AI classification → tool selection → tier selection → domain | — |
|
||
| 11.5 Website: AI business classifier | 1d | Gemini Flash integration for business type classification | — |
|
||
| 11.6 Website: resource calculator | 0.5d | Live RAM/disk calculation based on selected tools | — |
|
||
|
||
#### Week 12: End-to-End Integration
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 12.1 Website: payment flow (Stripe Checkout) | 1d | Stripe integration, order creation | 11.4 |
|
||
| 12.2 Website: provisioning status page (SSE) | 1d | Real-time progress display | 11.1, 12.1 |
|
||
| 12.3 End-to-end test: payment → provision → AI ready → mobile chat | 3d | Full journey on staging VPS | All above |
|
||
| 12.4 Provisioner: Playwright scenario migration (7 scenarios, minus n8n) | 2d | Cal.com, Chatwoot, Keycloak, Nextcloud, Stalwart, Umami, Uptime Kuma via OpenClaw browser | 11.1 |
|
||
| 12.5 Mobile: settings screens (agent config, autonomy, external comms) | 1d | Agent management, model selection, external comms gate | 10.1 |
|
||
| 12.6 Mobile: secrets side-channel (provide/reveal) | 1d | Secure modal for credential input, tap-to-reveal card | Phase 2 |
|
||
|
||
### Phase 3 Exit Criteria
|
||
|
||
- [ ] Full customer journey works: website signup → payment → provisioning → AI ready
|
||
- [ ] Mobile app: login, chat with agents, approve commands, view usage
|
||
- [ ] Provisioner deploys OpenClaw + Safety Wrapper (not orchestrator/sysadmin)
|
||
- [ ] n8n references fully removed
|
||
- [ ] config.json no longer contains plaintext passwords
|
||
- [ ] Chat relay works: App → Hub → Safety Wrapper → OpenClaw → response
|
||
- [ ] Push notifications delivered for approval requests
|
||
|
||
---
|
||
|
||
## 5. Phase 4 — Polish & Launch (Weeks 13-16)
|
||
|
||
### Goal: Security audit, performance optimization, founding member launch
|
||
|
||
#### Week 13: Security Audit + P1 Adapters
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 13.1 Security audit: secrets redaction (adversarial testing) | 2d | Test with crafted payloads: encoded, nested, multi-format | Phase 3 |
|
||
| 13.2 Security audit: command gating (boundary testing) | 1d | Attempt to bypass classification via edge cases | Phase 3 |
|
||
| 13.3 Security audit: path traversal, injection, SSRF | 1d | Penetration testing of all Safety Wrapper endpoints | Phase 3 |
|
||
| 13.4 Run `openclaw security audit --deep` on staging | 0.5d | Fix any findings | Phase 3 |
|
||
| 13.5 Cheat sheets: Odoo, Listmonk, NocoDB, Umami, Keycloak, Activepieces | 3d | P1 tool adapters operational | — |
|
||
| 13.6 Channel configuration: WhatsApp + Telegram | 1.5d | OpenClaw channel config; pairing mode; DM security | — |
|
||
|
||
#### Week 14: Performance + Polish
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 14.1 Prompt caching optimization | 1d | Verify cacheRetention: "long" working; measure cache hit rate | Phase 3 |
|
||
| 14.2 Token efficiency audit | 1d | Measure per-agent token usage; optimize verbose SOUL.md files | 14.1 |
|
||
| 14.3 Secrets redaction performance benchmark | 0.5d | Confirm <10ms latency with 50+ secrets in registry | Phase 3 |
|
||
| 14.4 Mobile app: UI polish, error handling, offline state | 2d | Production-ready mobile experience | Phase 3 |
|
||
| 14.5 Website: remaining pages (agent config, payment, provisioning status) | 1.5d | Complete onboarding flow | Phase 3 |
|
||
| 14.6 Provisioner: integration tests (Docker Compose based) | 2d | Test provisioning in container; verify all steps succeed | Phase 3 |
|
||
|
||
#### Week 15: Staging Launch + First-Hour Templates
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 15.1 Deploy full stack to staging | 1d | Hub + Website + Provisioner + staging tenant VPS | All above |
|
||
| 15.2 Internal dogfooding: team uses staging for 1 week | 5d (ongoing) | Bug reports, UX feedback, performance data | 15.1 |
|
||
| 15.3 First-hour templates: Freelancer workflow | 1d | Email setup, calendar connect, basic automation | 15.1 |
|
||
| 15.4 First-hour templates: Agency workflow | 1d | Client comms, project tracking, team setup | 15.1 |
|
||
| 15.5 Backup monitoring via OpenClaw cron | 0.5d | Daily backup-status.json check + Hub reporting | 15.1 |
|
||
| 15.6 Interactive demo: ephemeral container system | 2d | Per-session demo with 15-min TTL | 15.1 |
|
||
|
||
#### Week 16: Launch
|
||
|
||
| Task | Effort | Deliverable | Depends On |
|
||
|------|--------|-------------|-----------|
|
||
| 16.1 Fix staging issues from dogfooding | 3d | All critical/high issues resolved | 15.2 |
|
||
| 16.2 Production deployment | 1d | Hub production, pre-provisioned server pool, DNS | 16.1 |
|
||
| 16.3 Founding member onboarding: first 10 customers | ongoing | Hands-on onboarding, 2× token allotment | 16.2 |
|
||
| 16.4 Monitoring dashboard setup | 0.5d | Hub health, tenant health, billing dashboards | 16.2 |
|
||
| 16.5 Runbook documentation | 0.5d | Incident response, common issues, escalation paths | 16.2 |
|
||
|
||
### Phase 4 Exit Criteria
|
||
|
||
- [ ] Security audit passes with no critical findings
|
||
- [ ] Performance targets met (redaction <10ms, heartbeat reliable, tool calls <5s p95)
|
||
- [ ] 10 founding members onboarded and actively using the platform
|
||
- [ ] WhatsApp and Telegram channels operational
|
||
- [ ] Interactive demo working on letsbe.biz/demo
|
||
- [ ] Backup monitoring reporting to Hub
|
||
- [ ] First-hour templates proving cross-tool workflows work
|
||
|
||
---
|
||
|
||
## 6. Dependency Graph
|
||
|
||
```
|
||
┌─────────────┐
|
||
│ 1.1 Monorepo│
|
||
│ Setup │
|
||
└──────┬──────┘
|
||
┌──────┴──────┐
|
||
┌─────┤ ├─────┐
|
||
│ │ │ │
|
||
┌──────▼──┐ ┌▼────────┐ ┌─▼──────────┐
|
||
│1.2 SW │ │1.3 SQLite│ │3.1 Secrets │
|
||
│Skeleton │ │Schema │ │Proxy Server│
|
||
└────┬────┘ └────┬────┘ └─────┬──────┘
|
||
│ │ │
|
||
┌────▼────┐ ┌────▼────┐ ┌───▼────────┐
|
||
│1.5 Tool │ │1.4 Secrets│ │3.2-3.5 │
|
||
│Execute │ │Registry │ │4-Layer │
|
||
│Endpoint │ └────┬─────┘ │Redaction │
|
||
└────┬────┘ │ └───┬────────┘
|
||
│ │ │
|
||
┌────▼────┐ │ ┌───▼────────┐
|
||
│2.1 Cmd │ │ │3.6 P0 Tests│
|
||
│Classify │ │ │Redaction │
|
||
└────┬────┘ │ └────────────┘
|
||
│ │
|
||
┌─────────┼─────┐ │
|
||
│ ┌────┤ │ │
|
||
│ │ │ │ │
|
||
┌─▼──┐┌▼──┐┌▼──┐ │ │
|
||
│2.2 ││2.3││2.4│ │ │
|
||
│Shell│Dock│File│ │ │
|
||
│Exec││er ││Exec│ │ │
|
||
└────┘└───┘└───┘ │ │
|
||
│ │
|
||
┌────▼─────▼──┐
|
||
│4.1 Autonomy │
|
||
│Engine │
|
||
└──────┬──────┘
|
||
│
|
||
┌──────▼──────┐
|
||
│4.4 OpenClaw │
|
||
│Integration │
|
||
└──────┬──────┘
|
||
│
|
||
┌─────────┼──────────┐
|
||
│ │ │
|
||
┌────▼───┐ ┌───▼────┐ ┌──▼─────────┐
|
||
│5.1-5.4 │ │6.1-6.7 │ │7.1-7.9 │
|
||
│Hub │ │Token │ │Tool │
|
||
│Protocol│ │Billing │ │Adapters │
|
||
└────┬───┘ └───┬────┘ └──┬─────────┘
|
||
│ │ │
|
||
┌────▼─────────▼─────────▼──┐
|
||
│8.1-8.7 Approvals + Config │
|
||
└────────────┬──────────────┘
|
||
│
|
||
┌────────────┼────────────┐
|
||
│ │ │
|
||
┌───▼────┐ ┌────▼───┐ ┌──────▼──────┐
|
||
│9.1-9.6 │ │10.1-10.5│ │11.1-11.6 │
|
||
│Mobile │ │Customer│ │Provisioner │
|
||
│App │ │Portal │ │+ Website │
|
||
└───┬────┘ └───┬────┘ └──────┬──────┘
|
||
│ │ │
|
||
└──────────┼─────────────┘
|
||
│
|
||
┌──────────▼──────────┐
|
||
│12.3 E2E Integration │
|
||
└──────────┬──────────┘
|
||
│
|
||
┌──────────▼──────────┐
|
||
│Phase 4: Polish │
|
||
│Security + Launch │
|
||
└─────────────────────┘
|
||
```
|
||
|
||
---
|
||
|
||
## 7. Parallel Workstreams
|
||
|
||
Tasks that can be developed simultaneously by different engineers:
|
||
|
||
### Stream A: Safety Wrapper Core (1 senior engineer)
|
||
```
|
||
Week 1-2: SW skeleton, classification, executors
|
||
Week 3: Autonomy engine, SECRET_REF injection
|
||
Week 4: OpenClaw integration, integration tests
|
||
Week 5-6: Hub client, heartbeat, config sync
|
||
Week 7-8: Token metering, approval round-trip
|
||
```
|
||
|
||
### Stream B: Secrets Proxy (1 engineer)
|
||
```
|
||
Week 1-2: Proxy skeleton, 4-layer pipeline
|
||
Week 3: P0 tests (TDD), performance benchmarks
|
||
Week 4: Integration with OpenClaw LLM routing
|
||
Week 5+: Secrets API (provide/reveal/generate/rotate)
|
||
```
|
||
|
||
### Stream C: Hub Backend (1 engineer)
|
||
```
|
||
Week 1-4: Prisma models, tenant API endpoints
|
||
Week 5-6: Billing pipeline, Stripe meters
|
||
Week 7-8: Approval queue, agent config CRUD
|
||
Week 9-10: Customer portal API, chat relay
|
||
```
|
||
|
||
### Stream D: Mobile + Frontend (1 engineer)
|
||
```
|
||
Week 1-4: (Can start UI mockups, design system)
|
||
Week 5-8: (Website landing page, onboarding flow)
|
||
Week 9-10: Mobile app core (auth, chat, approvals)
|
||
Week 11-12: Polish, settings, usage dashboard
|
||
```
|
||
|
||
### Stream E: Provisioner + DevOps (1 engineer, part-time)
|
||
```
|
||
Week 1-4: Docker image builds, CI/CD pipeline
|
||
Week 5-8: Tool cheat sheets (P0 + P1)
|
||
Week 9-11: Provisioner update, n8n cleanup
|
||
Week 12: Integration testing, config.json fix
|
||
```
|
||
|
||
**Minimum team size: 3 engineers** (streams A+B combined, C, D+E combined)
|
||
**Recommended team size: 4-5 engineers** (each stream dedicated)
|
||
|
||
---
|
||
|
||
## 8. Scope Cut Table
|
||
|
||
If timeline pressure hits, these items can be deferred to post-launch:
|
||
|
||
| Item | Phase | Impact of Deferral | Difficulty to Add Later |
|
||
|------|-------|-------------------|------------------------|
|
||
| Interactive demo | 4 | No demo on website — use video instead | Low |
|
||
| WhatsApp/Telegram channels | 4 | App-only access — channels are config, not code | Low |
|
||
| P2+P3 tool cheat sheets | 4 | 6 tools instead of 24 at launch | Low |
|
||
| DNS automation | 3 | Manual DNS record creation (existing flow) | Low |
|
||
| First-hour workflow templates | 4 | No guided first hour — users explore freely | Low |
|
||
| Customer portal web UI | 3 | Mobile app only — no web dashboard for customers | Medium |
|
||
| Overage billing | 2 | Pause AI at pool limit (no overage option) | Medium |
|
||
| Custom agent creation | 3 | 5 default agents only, no custom | Medium |
|
||
| Founding member program | 2 | Standard pricing only — add multiplier later | Low |
|
||
| Dynamic tool installation | Post-launch | Fixed tool set per provisioning — no add/remove | Medium |
|
||
| Premium model tier | 2 | Included models only — add premium later | Medium |
|
||
|
||
### Non-Negotiable (Cannot Cut)
|
||
|
||
- Secrets redaction (the privacy guarantee)
|
||
- Command classification + gating
|
||
- Hub ↔ Safety Wrapper communication
|
||
- Token metering (needed for billing even without overage)
|
||
- Mobile app (primary customer interface)
|
||
- Provisioner update (must deploy new stack)
|
||
- 6 P0 tool cheat sheets
|
||
|
||
---
|
||
|
||
## 9. Critical Path
|
||
|
||
The longest chain of dependent tasks that determines the minimum project duration:
|
||
|
||
```
|
||
Monorepo setup (2d)
|
||
→ Safety Wrapper skeleton (2d)
|
||
→ Command classification (3d)
|
||
→ Executors (2d)
|
||
→ Autonomy engine (2d)
|
||
→ OpenClaw integration (2d)
|
||
→ Hub protocol (5d)
|
||
→ Token metering + billing (5d)
|
||
→ Approval queue (4d)
|
||
→ Customer portal API (3d)
|
||
→ Chat relay (2d)
|
||
→ Mobile app chat (3d)
|
||
→ Provisioner update (3d)
|
||
→ E2E integration test (3d)
|
||
→ Security audit (3d)
|
||
→ Launch (1d)
|
||
|
||
Total critical path: ~42 working days ≈ 8.5 weeks
|
||
```
|
||
|
||
With parallelization (5 engineers), the 16-week timeline has ~7.5 weeks of buffer distributed across phases. This buffer absorbs:
|
||
- Unexpected OpenClaw integration issues
|
||
- Secrets redaction edge cases requiring additional work
|
||
- Mobile app platform-specific bugs (iOS/Android)
|
||
- Provisioner testing on real VPS hardware
|
||
|
||
---
|
||
|
||
*End of Document — 04 Implementation Plan*
|