## Weekly Newsletter (Jan 25–31, 2026)

### 1) Executive Summary

This week was about **making ElizaOS easier to start, safer to operate, and more “real-world capable.”** Three standout milestones:

- **Onboarding unblocked:** The team fixed a critical failure in the `elizaos create` flow that prevented new developers from generating projects, and updated docs to reflect the correct CLI install path. That “front door” is open again.  
- **Agents leveled up for production workflows:** Work accelerated on **Eliza v2 + automation**, including early **n8n workflow engine integration** planning and continued improvements to runtimes, streaming, and plugin ergonomics.  
- **Ecosystem infrastructure strengthened:** Security and analytics got meaningful upgrades—formal vulnerability reporting was added, and the org-wide repo health tracking/analytics effort expanded.

---

### 2) Development Updates

#### Core framework & developer experience
- **CLI project generation fix (critical):** The “cannot generate project” issue was resolved across code + docs, restoring a smooth quickstart path for newcomers. See issue **[#6388](https://github.com/elizaos/eliza/issues/6388)** and PR **[#6389](https://github.com/elizaos/eliza/pull/6389)**.  
- **Security process formalized:** A `SECURITY.md` was added to standardize vulnerability reporting and response (**[#6428](https://github.com/elizaos/eliza/pull/6428)**).  
- **Runtime performance & reliability improvements:** January work continued to land, including optimizations to initialization and multi-step workflows (e.g., provider handling, retry logic, and streaming support in various parts of the stack as reflected in the week’s repo activity summary).

#### Automation: n8n as the primary workflow layer
A major technical direction sharpened in discussion: **use n8n as the integration backbone** instead of building one-off plugins for every SaaS.

- The open issue tracking this is **[“Integrate N8N Workflow Engine” #6429](https://github.com/elizaos/eliza/issues/6429)**.  
- Proposed architecture (as discussed in Discord) is a **three-layer stack**:  
  1) **AI Workflow Generator** (NL → workflow JSON)  
  2) **Workflow Engine** (TaskService/cron execution)  
  3) **OAuth Gateway** (multi-tenant credentials)

This aligns with the belief that “most users want workflow automation (email/calendar/notion/linear + conditional logic), not full autocoding.”

#### Knowledge & embeddings
- Community debugging identified two recurring themes:
  - **Embedding provider confusion:** Some users attempted to use `openrouter` where the enum only accepts `openai | google` in certain configurations; guidance was to use `openai` as embedding provider (and separately choose models via OpenRouter where applicable).  
  - **Cost blowups from contextual embeddings:** Enabling contextual embeddings (`CTX_KNOWLEDGE_ENABLED=true`) can balloon token usage and cost. Several members shared minimal configs to keep costs predictable.

Also noted: plans to **move embeddings fully into plugins in v2.x**, which should make provider behavior clearer and easier to swap.

#### Social integration & auth
- The core-dev thread focused on a **“connection page”** flow: users link social accounts, then get redirected back to the bot/session.  
- The team is evaluating **Composio** for in-chat auth and 100+ integrations—and importantly, **Stan already built a Composio plugin months ago**:  
  - Repo: **https://github.com/standujar/plugin-composio**  
  - Stan also mentioned an upcoming RFC with implementation ideas.

#### Critical bug: callback ordering in Eliza 1.7.2
Victor Creed reported a **high-impact bug** affecting custom plugins using `plugin-sql`, `plugin-openai`, and `plugin-bootstrap`: action callbacks execute in **reverse order**, and a documented “structured return” message can be **omitted entirely**. This remains unresolved and is a priority candidate for a targeted patch.

---

### 3) Community Spotlight (Discord)

- **Routing research: “ElizaRouter” concept:** DorianD led a thoughtful exploration of model routing frameworks (RouteLLM, Latitude.so, MasRouter) and proposed a distributed router where receiving nodes select optimal nodes with **time-decay QoS**. Candidate small models included **Phi-4 Mini, Qwen2.5/Qwen3 (1.5B–7B), Mistral 7B, DeepSeek-Coder**. The conversation was equal parts pragmatic (cost/load) and architectural (distributed selection strategy).  
- **Dev-support wins:**  
  - Odilitime helped troubleshoot a `@elizaos/plugin-web-search` “Cannot find module” install issue by recommending package.json resolution tweaks and trying `bun`.  
  - Chucknorris recommended switching from **SSE to socket.io** to resolve MIME type streaming issues in certain deployments.  
- **Docs-as-code momentum:** sayonara encouraged PR-based docs contributions via **https://github.com/elizaOS/docs**, and the community continued aligning around treating documentation like “game manuals” that improve agent performance.

---

### 4) Token Economics (AI16z → ELIZAOS, and auto.fun)

#### Migration & utility discussion (high attention this week)
Token conversation was intense and candid:

- **Allocation and trust concerns:** Community members questioned the post-migration distribution—especially the perceived **40% team allocation** after a **1:10 increase**, and asked for clarity on vesting and known wallets (questions remain open).  
- **Migration UX + liquidity pain:** At least one user reported **~98% loss** attempting to swap ETH-chain ELIZAOS back to ETH/USDT due to **zero liquidity**, highlighting the need for clearer official guidance and safer migration paths.  
- **Utility clarified by the team:** Odilitime reiterated that ELIZAOS is intended to be **the accepted currency in products**, including **gas for Jeju**, with planned **credit card payment rails feeding buybacks**.

#### auto.fun
No concrete auto.fun shipping updates appeared in this week’s captured notes. If you’re building around auto.fun, this is a good moment to post a short status request (or PR a “current state” doc) so the community can align expectations.

---

### 5) Coming Soon

Here’s what to watch next (and where help is welcome):

- **n8n workflow automation** moving from planning → working cloud execution (see **[#6429](https://github.com/elizaos/eliza/issues/6429)**).  
- **Social account connection page + in-chat auth**, likely leveraging Composio (review Stan’s plugin and RFC when it drops).  
- **Embedding architecture changes in v2.x** (embeddings moved fully into plugins).  
- **ERC-8004 mainnet launch (agent identity/reputation):** announced as imminent; keep an eye out for ecosystem tooling and tutorials once live.  
- **Fixes & stability:**
  - Action callback order bug in **1.7.2** (needs reproduction + patch).  
  - `@elizaos/plugin-web-search` module resolution/install ergonomics.  
  - Clearer migration instructions (especially ETH-chain edge cases).

---

### 6) Resources

**Key GitHub items**
- CLI “create” bug: **[#6388](https://github.com/elizaos/eliza/issues/6388)**, fix in **[#6389](https://github.com/elizaos/eliza/pull/6389)**  
- n8n integration tracker: **[#6429](https://github.com/elizaos/eliza/issues/6429)**  
- Security policy added: **[#6428](https://github.com/elizaos/eliza/pull/6428)**  
- Docs install fix (CLI scoping): **https://github.com/elizaOS/docs/pull/83**  
- Twitter plugin broker auth: **https://github.com/elizaos-plugins/plugin-twitter/pull/47**  
- Org analytics / repo tracking updates: **https://github.com/elizaos/elizaos.github.io/pull/238**

**Discord threads worth revisiting**
- AI routing + callback bug (coders):  
  https://discord.com/channels/1253563208833433701/1300025221834739744  
- Social auth + Composio (core-devs):  
  https://discord.com/channels/1253563208833433701/1377726087789940836  
- Token discussion + migration issues:  
  https://discord.com/channels/1253563208833433701/1253563209462448241  

**Community-built integration**
- Stan’s Composio plugin: **https://github.com/standujar/plugin-composio**