# ElizaOS Weekly Newsletter (2026-01-19 → 2026-01-25) ## 1) Executive Summary **This week was all about scaling capability while keeping the ecosystem coherent and safe.** Here are the big milestones: 1. **MCP + Eliza Cloud got a concrete “how-to” path.** We confirmed that *Eliza Cloud agents can use external Model Context Protocol (MCP) servers* by registering them via the Cloud API (`POST /api/v1/mcps`). That unlocks real integrations (and Red Teaming workflows) without waiting for first-party hosting. 2. **V2 groundwork continued: more resilient runtimes, better agent discovery, and major stability work.** The ongoing V2 direction (dynamic execution, Python/Rust parity work, ecosystem plugin sync) continues to mature, alongside important reliability improvements across core + plugins. 3. **Tokenomics and trust were front-and-center.** The community debated ecosystem token launches (notably Hyperscape/GOLD), while the team reiterated runway constraints and near-term product deliverables. In parallel, multiple *active scams* (fake “support” + wallet/migration requests) were identified—community vigilance helped prevent damage. --- ## 2) Development Updates (Core, Plugins, Infrastructure) ### MCP integration with Eliza Cloud (practical API clarity) A key technical highlight from Discord: **external MCP servers can be registered in Eliza Cloud** using: - `POST /api/v1/mcps` → once registered, **Eliza Cloud can proxy calls** to those MCP endpoints for agent reasoning + task execution. This is especially relevant for advanced agent patterns like **AI Red Teaming with human-in-the-loop**, where tooling often lives outside the main agent runtime. ### Plugin ecosystem stability: Discord plugin fixes in the wild Several users hit real upgrade friction moving from **ElizaOS 1.6.x → 1.7.x**, including SQL migration failures and Discord bootstrap issues. Community troubleshooting confirmed that: - **ElizaOS 1.7.2 + Discord plugin 1.3.8** resolves key bootstrap errors. - Fix path for stubborn CLI version caching included clearing bun cache and reinstalling the CLI globally: - `bun pm cache rm` - `bun uninstall -g @elizaos/cli` - `bun i -g @elizaos/cli` - remove `node_modules` + lockfiles, and ensure `package.json` references the right version. ### Memory systems surfaced for future agent quality In `core-devs`, the team shared two projects worth watching if you’re building longer-lived agents: - **supermemory** (super fast memory engine) - **hindsight** (agent memory inspired by human memory patterns) These aren’t “official defaults” yet, but they’re strong signals of where agent reliability and personalization are heading. ### Infra & self-hosting patterns: mini PCs + free-tier cloud A recurring pattern: community members are running Eliza on **mini PCs** to avoid keeping main machines on 24/7. There was also discussion of an **Oracle Cloud free tier** deployment option (noted spec: **4 vCPU / 24GB RAM**), which is a surprisingly capable baseline for lightweight agent hosting. --- ## 3) Community Spotlight (Notable Contributions & Discussions) ### SOLOMON VANDY unblocked MCP integration for builders Big shoutout to **SOLOMON VANDY** for giving SATA a clear, actionable MCP integration answer—*including the exact endpoint* and the proxying behavior. This is the kind of “specificity” that turns a question into a shipped integration. ### 0xbbjoker’s upgrade playbook saved dev hours **0xbbjoker** continued to be the go-to for “why is my install haunted?” debugging—especially for bun/CLI caching problems and Discord plugin upgrade paths. ### New builders and ecosystem energy - A developer from Japan introduced themselves (AI + full stack, 8 years) and offered support—great signal that ElizaOS is attracting experienced builders. - **ElizaBAO** posted dev hiring with creator rewards + Polymarket affiliate profit sharing opportunities, reflecting growing community-led commercialization efforts. ### “Why aren’t there more one-trick agents?”—a product question worth answering Skinny raised an important adoption critique: if the framework is strong, **why don’t we see more specialized single-purpose agents deployed publicly?** Possible culprits the community implied: - unclear “golden path” templates - missing discovery/marketplace dynamics - builders keeping use cases private for competitive reasons This is a productive prompt for docs, examples, and go-to-market. --- ## 4) Token Economics (AI16z, $ELIZAOS, auto.fun + ecosystem tokens) ### Hyperscape/GOLD debate: funding vs focus The community spent significant time debating the **GOLD token** launch tied to Hyperscape fundraising. Concerns included dilution of attention from the core token, reputational risk, and launch mechanics that looked unfavorable to retail participants. From the team side, **Shaw shared critical context**: the project sits around a **$20M market cap** with roughly **8 months runway**, and the team doesn’t have spare budget to fund every ecosystem project directly. Hyperscape’s near-term plan is a **2-week MVP schedule**, with broader cloud apps “imminent” and Babylon cited at **~375k users**. ### Scams targeting token holders (important) Multiple scam patterns were reported: - “Create a Ticket” bot asking for **wallet addresses** - DMs claiming you must manually **migrate tokens by sending them to a wallet** **Reminder:** No legitimate support process should require you to send tokens to “migrate.” If you see these, report immediately. ### auto.fun / token clarity gap Community members explicitly asked how various tokens relate to ElizaOS and whether they impact the core project. That question went unanswered this week—suggesting we need a clearer “ecosystem token map” doc (including where **auto.fun** fits into launches, creator tokens, and app funding). --- ## 5) Coming Soon (What to Watch Next) - **External MCP support documentation & examples.** Now that `/api/v1/mcps` is confirmed as the registration path, expect community demand for a step-by-step guide + reference server templates. - **Hyperscape MVP progress.** The stated 2-week MVP timeline implies we’ll see more concrete demos, gameplay loops, and (hopefully) clearer token boundaries soon. - **Better “single-purpose agent” packaging.** If the community wants one-trick agents everywhere, we likely need curated templates, deployment recipes, and distribution/discovery mechanisms. - **Enterprise Twitter API key exploration.** Consultant Vivek recommended obtaining enterprise access—this could impact social agents, monitoring, and distribution tooling. --- ## 6) Resources (Key Links & Threads) ### Discord discussions - MCP + external servers (coders): https://discord.com/channels/1253563208833433701/1300025221834739744 - Core-devs memory links + Vivek note: https://discord.com/channels/1253563208833433701/1377726087789940836 - General discussion (tokens, adoption questions): https://discord.com/channels/1253563208833433701/1253563209462448241 ### GitHub / engineering references - Eliza V2 weekly summary (Jan 18–24): (see internal weekly digest referenced in community knowledge base) - ElizaOS core repo: https://github.com/elizaos/eliza ### Community-shared tooling & repos - DaVinci Resolve MCP integration (community project): https://github.com/AyeRubz/davinci-resolve-mcp - supermemory: https://github.com/supermemoryai/supermemory - hindsight: https://github.com/vectorize-io/hindsight ### Practical reminders - If a “support bot” asks for a wallet address or asks you to send tokens: **assume scam** and report. - If you’re integrating MCP into Cloud agents: track `/api/v1/mcps` and push for docs so we can standardize best practices.