molecule-ai · Apr 23, 2026 · Apr 23, 2026 · Apr 23, 2026 · Apr 23, 2026 · Apr 23, 2026
diff --git a/docs/blog/2026-04-20-chrome-devtools-mcp-seo/index.md b/docs/blog/2026-04-20-chrome-devtools-mcp-seo/index.md
@@ -4,6 +4,7 @@ date: 2026-04-20
 slug: browser-automation-ai-agents-mcp
 description: "Learn how to add browser automation to your AI agents using Chrome DevTools and the Model Context Protocol. Full Python code examples — no Puppeteer wrappers, no SaaS dependencies."
 tags: [MCP, browser-automation, AI-agents, CDP, tutorial]
+og_image: /assets/blog/2026-04-21-chrome-devtools-mcp-og.png
 ---
 
 # Give Your AI Agent a Real Browser: MCP + Chrome DevTools

diff --git a/docs/blog/2026-04-20-chrome-devtools-mcp/index.md b/docs/blog/2026-04-20-chrome-devtools-mcp/index.md
@@ -4,6 +4,7 @@ date: 2026-04-20
 slug: chrome-devtools-mcp
 description: "Chrome DevTools MCP gives any compatible AI agent full browser control through a standards-based interface. That's powerful for prototypes. For production, you need a governance layer. Here's where Molecule AI fits in."
 tags: [browser-automation, mcp, governance, chrome-devtools, security]
+og_image: /assets/blog/2026-04-21-chrome-devtools-mcp-og.png
 ---
 
 # Browser Automation Meets Production Standards

diff --git a/docs/blog/2026-04-21-cloudflare-artifacts/index.md b/docs/blog/2026-04-21-cloudflare-artifacts/index.md
@@ -4,6 +4,7 @@ date: 2026-04-21
 slug: cloudflare-artifacts-molecule-ai
 description: "Attach a Cloudflare Artifacts git repository to any Molecule AI workspace. Import existing repos, fork for experiments, mint short-lived git credentials — all via the platform API. Git-native storage for AI agents."
 tags: [Cloudflare, git, artifacts, AI-agents, workflow, tutorial]
+status: "⚠️ PM ruling 2026-04-22: sub-100ms claim UNSUBSTANTIATED — replaced with 'low-latency' (same latency class)"
 ---
 
 # Give Your AI Agent a Git Repository: Molecule AI + Cloudflare Artifacts
@@ -25,7 +26,7 @@ Git-native storage is different because:
 - **Agents already know git.** Clone, branch, commit, push. No new primitives to learn.
 - **Versioning is structural.** Every change is a commit. Rollback is `git revert`. No "last writer wins" data loss.
 - **Collaboration is native.** Fork a repo, experiment, open a PR. The same workflow humans use to collaborate applies to agents.
-- **Cloudflare Artifacts is fast.** Git operations run on Cloudflare's edge — sub-100ms clone times from anywhere. No S3 bandwidth bills.
+- **Cloudflare Artifacts is fast.** Git operations run on Cloudflare's edge — low-latency clone times from anywhere. No S3 bandwidth bills.
 - **Access control is git-native.** Token scoping, branch protection, repo-level permissions. The same model your team already uses.
 
 ## API Reference

diff --git a/docs/blog/2026-04-22-a2a-v1-agent-platform/index.md b/docs/blog/2026-04-22-a2a-v1-agent-platform/index.md
@@ -3,6 +3,8 @@ title: "What A2A v1.0 Means for Your Agent Stack: Why Protocol-Native Beats Prot
 description: "A2A v1.0 shipped March 2026 as the Linux Foundation's standard for multi-agent communication. Here's why being built on it from day one matters more than adding it as a layer."
 date: 2026-04-22
 canonical: https://docs.molecule.ai/blog/a2a-v1-agent-platform
+tags: [a2a, agent-protocol, multi-agent, governance, enterprise, platform]
+og_image: /assets/blog/2026-04-21-chrome-devtools-mcp-og.png
 ---
 
 *Meta description (160 chars): Before you buy an agent platform, ask how A2A delegation is attributed. The answer reveals everything about governance.*
@@ -49,7 +51,7 @@ Both Python and Node.js external agent templates are under 100 lines. Registrati
 
 ## Why This Matters Now: The Governance Gap in Competing Implementations
 
-A2A v1.0 ratification has accelerated adoption across the agent platform landscape. LangGraph shipped A2A support in Q1 2026 (PRs #6645, #7113 — still in review after 3+ months). But a protocol implementation and a governance-ready implementation are not the same thing.
+A2A v1.0 ratification has accelerated adoption across the agent platform landscape. LangGraph's A2A implementation (PRs #6645, #7113 — ⚠️ VERIFY: PMM 2026-04-21 confirmed these PRs not found in langchain-ai/langgraph open PR list; may be merged, closed, or re-numbered) positions against the governance gap. But a protocol implementation and a governance-ready implementation are not the same thing.
 
 LangGraph's current A2A PRs implement the protocol layer: message framing, capability negotiation, task routing. What they do not yet implement is the governance layer — the mechanisms that make A2A usable in regulated environments, multi-tenant deployments, and enterprise fleets.
 

diff --git a/docs/blog/2026-04-22-cloudflare-tunnel-migration/index.md b/docs/blog/2026-04-22-cloudflare-tunnel-migration/index.md
@@ -4,6 +4,7 @@ date: 2026-04-22
 slug: cloudflare-tunnel-migration
 description: "Phase 33 replaces Cloudflare Tunnel with direct-connect agent workspaces that get their own public IPs. Here's what changed, why, and what it means for your deployment."
 tags: [platform, infrastructure, cloud, deployment]
+og_image: /assets/blog/2026-04-21-chrome-devtools-mcp-og.png
 ---
 
 # Phase 33: From Cloudflare Tunnel to Direct Connect — How Molecule AI Agent Workspaces Get Their Own IP

diff --git a/docs/marketing/battlecard/phase-34-partner-api-keys-battlecard.md b/docs/marketing/battlecard/phase-34-partner-api-keys-battlecard.md
@@ -112,4 +112,21 @@ Phase 30 shipped `mol_ws_*` (per-workspace auth tokens). Phase 34 extends to `mo
 
 ---
 
-*PMM draft 2026-04-22 — Marketing Lead 2026-04-23 v2: (1) lead claim updated to verified "first-mover" language per Research team competitive audit (LangGraph Cloud, CrewAI, Azure AI Foundry, Dify, Flowise, n8n — no equivalent `mol_pk_*` found), (2) Phase 30 cross-sell updated to "first agent platform with both" framing, (3) Language to Avoid section resolved. GA DATE CONFIRMED: April 30, 2026. Still awaiting PM input on partner tiers and marketplace billing.*
+*PMM draft 2026-04-22 — Marketing Lead 2026-04-23 v2: (1) lead claim updated to verified "first-mover" language per Research team competitive audit (LangGraph Cloud, CrewAI, Azure AI Foundry, Dify, Flowise, n8n — no equivalent `mol_pk_*` found), (2) Phase 30 cross-sell updated to "first agent platform with both" framing, (3) Language to Avoid section resolved. GA DATE CONFIRMED: April 30, 2026. Still awaiting PM input on partner tiers and marketplace billing.*
+
+---
+
+## Marketing Lead Review — 2026-04-23
+
+**Status: APPROVED for Sales distribution and launch prep. Two action items below.**
+
+✅ **Lead claim** — "first-mover" framing is correct and Research-verified. Sales team can use this now.
+✅ **Phase 30 linkage** — cross-sell claim ("first agent platform with both layered token scoping and a first-class partner provisioning API") is clean and approved.
+✅ **GA Date** — April 30, 2026 confirmed. Update Triggers table: mark "Phase 34 GA date confirmed" as ✅ DONE.
+✅ **Competitive table** — accurate as of 2026-04-23. Monitor CrewAI and Azure AI Foundry monthly.
+
+⏳ **Action (PM):** Marketplace-native billing rows (AWS/GCP) show "⏳ PM to confirm" — need PM input before Sales uses those rows in enterprise deals. Flagged to PM via issue #1122 routing.
+
+⚠️ **Action (DevRel):** Partner onboarding guide and CI/CD example are marked "⏳ DevRel in progress" — must ship by April 28 to support April 30 launch. Verify DevRel ETA.
+
+**Ready to distribute to:** Sales team, partner AEs. Do NOT share marketplace billing rows externally until PM confirms.
diff --git a/docs/marketing/blog/2026-04-21-cloudflare-artifacts-integration.md b/docs/marketing/blog/2026-04-21-cloudflare-artifacts-integration.md
@@ -2,7 +2,7 @@
 
 **Source:** PR #641 (feat(platform): Cloudflare Artifacts demo integration #595), merged 2026-04-17
 **Issue:** #1174
-**Status:** Draft v1
+**Status:** REVIEWED — Marketing Lead 2026-04-23
 
 ---
 
@@ -22,7 +22,7 @@ Key properties:
 - **Versioned** — every snapshot is a Git commit, accessible and diffable
 - **Branching** — agents can fork an isolated copy before experimental changes
 - **Short-lived credentials** — Git tokens minted on demand, revoked automatically
-- **Edge-hosted** — CF's network means sub-50ms access from anywhere an agent runs
+- **Edge-hosted** — CF's network means low-latency access from anywhere an agent runs
 
 This is a first-mover integration. As of 2026-04-17, no other AI agent platform has shipped a Git-backed workspace snapshot feature. The [Cloudflare blog post](https://blog.cloudflare.com/artifacts-git-for-agents-beta/) has the full context.
 
@@ -94,3 +94,27 @@ Your agents stop being stateless. They become participants in a versioned, colla
 
 **Docs:** [Cloudflare Artifacts setup](/docs/guides/cloudflare-artifacts)
 **PR:** [PR #641 on GitHub](https://github.com/Molecule-AI/molecule-core/pull/641)
+
+---
+
+## Marketing Lead Review Notes (2026-04-23)
+- [x] "sub-50ms" claim removed → "low-latency" (same class as sub-100ms PM ruling 2026-04-22)
+- [x] No benchmarks or ship dates
+- [x] No competitor names
+- [x] No AgenticAI hashtags
+- [ ] Frontmatter missing: add date, slug, canonical, og_image before publish
+- [ ] Blog URL conflict: `docs/marketing/blog/` vs `docs/blog/2026-04-21-cloudflare-artifacts/` — consolidate to one location
+
+
+---
+
+## Marketing Lead Review Notes (2026-04-23)
+- [x] Self-review gate: clean (no unconfirmed dates, no person names, no benchmarks, no competitor disparagement)
+- [x] Social copy approved — ready for publish queue
+- [x] Demo script approved — ready for screencast recording
+- [x] Community QA package complete
+- [x] Path conflict resolved: canonical blog = `docs/blog/2026-04-21-cloudflare-artifacts/index.md` — publish from there
+- [x] TTS narration.mp3 exists at `marketing/demos/cloudflare-artifacts/narration.mp3` — include in video assembly
+- [ ] All workers offline (A2A delegation failing) — dispatch blocked until token rotation + A2A restored
+
+**Launch readiness: READY TO DISPATCH — awaiting A2A restore.**
diff --git a/docs/marketing/briefs/2026-04-22-partner-api-keys-positioning-brief.md b/docs/marketing/briefs/2026-04-22-partner-api-keys-positioning-brief.md
@@ -1,5 +1,5 @@
 # Phase 34: Partner API Keys — PMM Positioning Brief
-**Owner:** PMM | **Status:** Draft | **Date:** 2026-04-22
+**Owner:** PMM | **Status:** DRAFT (reviewed by Marketing Lead 2026-04-23) | **Date:** 2026-04-22
 **Assumptions:** GA date TBD (blocked on Phase 32 completion + infra); partner tiers TBD with PM
 
 ---
@@ -57,7 +57,7 @@ Phase 34 (Partner API Keys) ships a `mol_pk_*` scoped key type that lets CI/CD p
 **Solution:** Partner API Keys enable fully automated provisioning through marketplace billing APIs. A buyer clicks "Deploy on [Marketplace]", the marketplace calls the Partner API to provision an org, charges begin on the marketplace invoice, and the buyer lands in a fully configured dashboard.
 
 **Three claims:**
-1. **Automated provisioning end-to-end.** From click to running org in under 60 seconds — no manual handoff.
+1. **Automated provisioning end-to-end.** From click to running org — no manual handoff. ⚠️ Remove "under 60 seconds" — PM ruling 2026-04-22: unsubstantiated timing claims require a citable benchmark before use.
 2. **Marketplace-native billing.** Usage flows through the marketplace's existing invoicing, not a separate Molecule AI subscription.
 3. **API-first management.** Marketplaces manage orgs, seats, and deprovisioning via the same Partner API used for provisioning.
 
@@ -108,18 +108,22 @@ Phase 30 (Remote Workspaces) shipped the per-workspace auth token model (`mol_ws
 | CI/CD example (org lifecycle + test teardown) | DevRel | Not started |
 | Partner API Keys landing page section | Content Marketer | Not started |
 | Marketplace listing copy | Content Marketer | Not started |
-| Battlecard update (add Phase 34 row) | PMM | Not started |
+| Battlecard update (add Phase 34 row) | PMM | ✅ Done — staging `docs/marketing/battlecard/phase-34-partner-api-keys-battlecard.md` |
 | Partner tier pricing page | Marketing Lead / PM | TBD |
 
 ---
 
 ## Open Questions for PM / Marketing Lead
 
-1. Partner tiers: will there be multiple key tiers (e.g., `orgs:create` vs `orgs:manage` vs `orgs:delete`)? Pricing model?
-2. GA date: dependent on Phase 32 completion — any updated ETA?
-3. First design partner: is there a named partner in the pipeline we can use as a reference in the onboarding guide?
-4. Rate limits: what are the per-key rate limits? Do limits vary by tier?
-5. Key rotation: are partner keys rotatable, or is rotation a delete + recreate?
+| # | Question | Owner | Status | Notes |
+|---|----------|-------|--------|-------|
+| 1 | Partner tiers: multiple key tiers (`orgs:create` vs `orgs:manage` vs `orgs:delete`)? Pricing model? | PM | ⚠️ TBD — PM to answer | Draftable with placeholders; PM approval needed |
+| 2 | GA date: dependent on Phase 32 completion — any updated ETA? | PM | 🔴 P0 BLOCKER | Nothing moves without this |
+| 3 | First design partner: named partner in pipeline for onboarding guide reference? | PM | 🔴 P0 BLOCKER | Cannot finalize onboarding guide without this |
+| 4 | Rate limits: per-key limits? Do limits vary by tier? | PM | ⚠️ TBD — PM to answer | Draftable with placeholder TBD values |
+| 5 | Key rotation: rotatable, or delete + recreate? | PM | ⚠️ TBD — PM to answer | Delete + recreate is the current model; rotate is an enhancement |
+
+**2026-04-23 update (PMM):** Q1, Q4, Q5 are advanceable with TBD placeholders — PM only needs to confirm or adjust. Q2 (GA date) and Q3 (design partner) are genuine P0 blockers that require PM input.
 
 ---
 

diff --git a/docs/marketing/launches/phase-34-community-announcement.md b/docs/marketing/launches/phase-34-community-announcement.md
@@ -0,0 +1,83 @@
+# Phase 34 Community Announcement
+
+**Channel:** Discord `#announcements` + GitHub Discussions  
+**Date:** 2026-04-23  
+**Author:** Marketing Lead (drafted on behalf of Community Manager)
+
+---
+
+🚀 **Phase 34 shipped — and it's a big one.**
+
+This phase is all about giving platform builders the visibility, control, and provisioning primitives they've been asking for. Four features landed today:
+
+---
+
+## What's new
+
+### 🔍 Tool Trace — see exactly what your agents did
+Every A2A response now includes a `tool_trace` in `Message.metadata`. It's a list of every tool your agent called, with the input it sent and a preview of the output it got back:
+
+```json
+"tool_trace": [
+  { "tool_name": "web_search", "input": {"query": "molecule ai agent platform"}, "output_preview": "Molecule AI is a multi-agent..." },
+  { "tool_name": "write_file",  "input": {"path": "summary.md"}, "output_preview": "File written (412 bytes)" }
+]
+```
+
+Parallel tool calls are handled — start/end events are paired via `run_id`, so concurrent calls don't get mixed up. Capped at 200 entries to prevent runaway loops from bloating your logs.
+
+The full trace is stored in `activity_logs.tool_trace` — queryable, auditable, and there when you need to debug why an agent did what it did.
+
+**Why it matters:** No more guessing. When something goes wrong in a multi-agent workflow, you now have the full tool call history to diagnose it.
+
+---
+
+### ⚙️ Platform Instructions — system prompt for your whole org
+Org admins can now configure system-level instructions that apply across every agent in the org:
+
+```http
+PUT /cp/platform-instructions
+{ "instructions": "Always respond in English. Tag every response with the agent workspace ID." }
+```
+
+Set it once, and every agent in your org inherits it — without touching individual workspace configs. Great for compliance requirements, house-style rules, or shared context that all your agents need.
+
+---
+
+### 🔑 Partner API Keys (`mol_pk_*`) — GA April 30
+The partner provisioning API is entering GA on **April 30**. If you're building a platform on top of Molecule AI — a marketplace, a CI/CD integration, or a multi-tenant product — you can now programmatically create and manage Molecule AI orgs via API:
+
+```http
+POST /cp/admin/partner-keys
+DELETE /cp/admin/partner-keys/:id
+```
+
+Ephemeral test orgs per PR. Org-scoped keys that can't escape their boundary. Automated teardown that stops billing on `DELETE`. No browser session required.
+
+We believe this makes Molecule AI the first agent platform with a first-class partner provisioning API. If you're a platform builder and want early access ahead of April 30, drop a message in `#partner-program`.
+
+---
+
+### 🌐 SaaS Fed v2 — improved multi-org federation
+Federation for multi-org deployments got a round of improvements in this phase. If you're running federated Molecule AI across multiple orgs, check the changelog for the specifics — the headline is better reliability and cleaner org boundary enforcement.
+
+---
+
+## Quick start
+
+- **Tool Trace**: no config needed — it's in every A2A response today. Check `message.metadata.tool_trace`.
+- **Platform Instructions**: `PUT /cp/platform-instructions` with your org admin token.
+- **Partner API Keys**: docs at `/docs/api/partner-keys` — GA April 30.
+
+---
+
+## Try it & tell us what you think
+
+All four features are live now (Partner API Keys GA on April 30). Spin them up and let us know:
+
+- 🐛 Bugs or unexpected behavior → `#bug-reports`
+- 💡 Feature requests → `#feedback`  
+- 🤝 Partner program interest → `#partner-program`
+- ❓ Questions → reply here or open a GitHub Discussion
+
+We read everything. — The Molecule AI team