v3.4.0 — WhatsApp IP Isolation

What's New

WhatsApp IP Isolation for Multi-Tenant Deployments

Each tenant now gets a unique Cloudflare exit IP for WhatsApp, preventing cross-account flagging when running multiple agents on the same server.

# Isolate a tenant's WhatsApp IP (one command):
./deploy/ip-isolate.sh acme-corp

How it works:

tenant-a → wireproxy :40001 → WARP Account A → Cloudflare IP-A
tenant-b → wireproxy :40002 → WARP Account B → Cloudflare IP-B

• Free Cloudflare WARP account per tenant (no cost)
• wireproxy userspace proxy (~4MB RAM each)
• Auto-injects ALL_PROXY into Docker container env
• Supports both Docker multi-tenant and standalone modes
• Integrated into deploy.sh as step 8/8 (IP_ISOLATE=true)

New Files

File	Description
deploy/ip-isolate.sh	One-command IP isolation script
SKILL.md	Root skill manifest for ClawHub

Config Changes

# New options in config.sh:
IP_ISOLATE=true           # Auto-run ip-isolate after deploy
IP_SOCKS_PORT=""          # SOCKS5 port (auto-assigned if empty)

Full Changelog

• feat: deploy/ip-isolate.sh — per-tenant Cloudflare WARP proxy
• feat: Integrated IP isolation as deploy step 8/8
• chore: Added IP_ISOLATE, IP_SOCKS_PORT to config.sh.example
• docs: Architecture diagram and usage guide in README.md

v3.3.1 — OpenClaw 2026.4.1 Compatibility

What's Changed

Fixes compatibility issues introduced by OpenClaw 2026.4.1 update:

Bug Fixes

• Exec approval timeout: Deploy now writes exec-approvals.json with security=full defaults, preventing gateway approval timeouts that blocked all tool execution
• Telegram DM blocking: Changed default dmPolicy from "pairing" to "open" so new contacts can message the bot directly
• Stale gateway token: Re-deploys now run openclaw gateway install --force to refresh embedded tokens

Config Additions

• tools.profile: "full" added to generated openclaw.json
• allowFrom: ["*"] added to Telegram channel config
• exec-approvals.json auto-generated with full permissions during deploy

Upgrade Notes

For existing deployments, run:

# On your server:
openclaw gateway install --force
openclaw gateway restart

Or re-deploy with ./deploy.sh <client-name> to apply all fixes automatically.

v3.3.0 — Dual-Threshold Compression & Memory Ranking

What's New

Cherry-picked 3 ideas from OpenViking Memory Plugin (no new dependencies, no new services).

1. Dual-Threshold Compression

• 50% (BACKGROUND_SAVE): Non-blocking extraction of key facts to ChromaDB — protects critical data early
• 65% (COMPRESS): Full L2 compression via haiku-class model (existing behavior, unchanged)
• Backward-compatible: TOKEN_THRESHOLD export still works

2. Recency-Weighted Search Ranking

Replaced simple word-count scoring with 3-factor ranking:

• Lexical overlap (normalized 0-1) × 0.5
• Recency decay (30-day half-life, floor 0.5) × 0.3
• Tag boost (order +0.12, quote +0.10, commitment +0.10, objection +0.08)

Also fixed bug: score + 0.5 was a no-op (not assigned back).

3. Archive Expand Command

chroma:expand <turn_id>

View full original text of any compressed/archived turn by ID.

Config Change

• softThresholdTokens: 12000 → 10000 (earlier flush hook trigger for dual-threshold)

Files Changed

File	Change
scripts/proactive-summary.mjs	Dual-threshold logic
skills/chroma-memory/chroma.mjs	Ranking + expand + bugfix
workspace/MEMORY.md	Updated architecture docs

v3.2.0 — Telegram Best Practices & Market-Adaptive Channels

What's New in v3.2.0

Telegram Upgraded to Strategic Channel

Telegram is no longer a "secondary channel" — it's now a full SDR-capable channel with unique advantages over WhatsApp.

New: skills/telegram-toolkit/

Feature	Description
Bot Commands	/catalog, /quote, /status, /contact, /language — structured self-service
Inline Keyboards	One-tap BANT qualification (product → volume → timeline), 3-5x faster than free-text
Large File Routing	2GB limit vs WhatsApp's 50MB — catalogs, certifications, video demos
Nurture Cadence	Full Telegram nurture sequence (Day 0 → Day 60+) with no window restrictions
Channel Broadcast	One-to-many product announcements via Telegram Channel

Market-Adaptive Channel Priority

Channel priority now depends on the customer's market — no longer hardcoded:

Market	Primary	Secondary
Africa / Latin America / South Asia	WhatsApp	Email
Middle East / Southeast Asia	WhatsApp	Telegram
Russia / CIS / Iran / Eastern Europe	Telegram	Email
Europe / Turkey	WhatsApp	Telegram

WhatsApp 72h Window Auto-Switch

New HEARTBEAT check (#13):

• 48h warning: Send follow-up on WhatsApp before window expires
• 72h expired: Auto-switch to Telegram (no window limit) or Email
• Never mark CRM as "contacted" if delivery actually failed

Other Changes

• WhatsApp corrected to "Business App" (not API) throughout documentation
• TOOLS.md Telegram section expanded from 2 lines to comprehensive guide
• README updated: 13 cron jobs, 7 pre-built skills
• All changes synced to openclaw-delivery/_template

Full Changelog

• a42980c feat: Telegram best practices — market-adaptive channels, bot commands, inline keyboards
• 721f120 feat: upgrade to 4-layer anti-amnesia architecture (v2.0)
• 92e6aef feat: wire anti-amnesia operating protocol into workspace MD files

---

Telegram Quick Setup:

# 1. Get bot token from @BotFather
# 2. Edit config
TELEGRAM_ENABLED=true
TELEGRAM_BOT_TOKEN="your-token"
# 3. Deploy — done. Full SDR pipeline works on Telegram.

v3.1.0 — 4-Layer Anti-Amnesia Architecture

What's New in v3.1.0

4-Layer Anti-Amnesia System (v2.0)

Your AI SDR now has a complete 4-layer memory system — it never forgets a customer, a quote, or a commitment.

Message In → L1 MemOS auto-recall (structured memory injection)
           → L3 ChromaDB per-turn store (customer-isolated, auto-tagged)
           → L2 Proactive Summary at 65% tokens (haiku compression)
           → L4 CRM Snapshot daily 12:00 (disaster recovery fallback)

Layer	Engine	What It Does
L1: MemOS	Structured memory	Auto-extracts BANT, commitments, objections every turn
L2: Proactive Summary	Token monitoring	Compresses at 65% context usage, all numbers preserved verbatim
L3: ChromaDB	Per-turn vector store	Every turn stored with customer_id isolation + auto-tags
L4: CRM Snapshot	Daily backup	Pipeline state backup as disaster recovery fallback

New Files

• skills/chroma-memory/ — Per-turn conversation storage with customer isolation, auto-tagging (quotes, commitments, objections, orders, samples), and CRM snapshot capability
• scripts/proactive-summary.mjs — Token usage monitoring + haiku-class compression trigger

Updated Workspace (7-Layer Context System)

• MEMORY.md — Full 4-layer operating protocol with data flow diagram, command reference, and priority matrix
• SOUL.md — 7-rule Memory Protocol (up from 5) with ChromaDB integration
• AGENTS.md — 4-layer memory management embedded in sales workflow
• HEARTBEAT.md — Now 12 automated checks (added Memory Health + CRM Snapshot)
• TOOLS.md — ChromaDB tool documentation added
• ANTI-AMNESIA.md — Upgraded to v2.0 with L4 chapter and 4-layer integration flow

Deployment Updates

• chromadb added to core skill profile (auto-installed)
• config.sh.example now includes ChromaDB configuration options
• Compatible with OpenClaw container-per-tenant deployment

Full Changelog

• 721f120 feat: upgrade to 4-layer anti-amnesia architecture (v2.0)
• 92e6aef feat: wire anti-amnesia operating protocol into workspace MD files
• 624af79 feat: add 3-layer Anti-Amnesia system for persistent customer memory
• 83406f7 security: comprehensive hardening based on production audit
• 6585c97 feat: open dmPolicy by default + control dashboard access
• 3bf6b0e feat: upgrade to 10-stage pipeline with full automation
• ef81d89 fix: add missing CRM Entry stage to all 8 README translations
• cfbf869 docs: add 8-language README support
• 4f57b92 Add ClawHub one-command install + fix git clone URLs
• c75a858 Initial release: B2B SDR Agent Template

---

Quick Start:

# OpenClaw users
clawhub install b2b-sdr-agent

# Full deployment
git clone https://github.com/iPythoning/b2b-sdr-agent-template.git
cd b2b-sdr-agent-template/deploy
cp config.sh.example config.sh
# Edit config.sh, then:
./deploy.sh my-company