# ElizaOS Weekly Newsletter (2026-05-14) *Week of **2026-05-08** to **2026-05-14*** ## 1) Executive Summary **1) Eliza-1 model lineup expanded (Qwen3.5-based).** This week, **zadayos** shared the latest **Eliza-1** series: a practical spectrum from **0.6B** (offline-capable on phones) through **27B-1M** (with **~1M token context**), plus a **0.8B** model currently training on **H200** hardware. The big takeaway: the project is pushing for *deployment-appropriate* models—from mobile-first agents to long-context server workloads. **2) auto.fun AI Sprint wrapped with winners announced.** After weeks of building, voting, and live demos, the **auto.fun AI Sprint** posted results: **Build4** took 1st, **Clawdyland** 2nd (with a 3rd place also mentioned). This is a strong signal that the ecosystem is not only shipping core tech, but also cultivating “builders who ship.” **3) Integration momentum: GODL proposes an agent-native “skill.md” pathway.** In the coders channel, **blankey1717** proposed integrating **GODL** (a gamified on-chain mining protocol on Solana) with elizaOS using SDKs + websockets + a public **skill.md**. **odilitime** confirmed that elizaOS can consume **skill.md as-is**, making this a low-friction candidate for real agent competitions, wallet-managed strategies, and on-chain automation experiments. --- ## 2) Development Updates (Core + Plugins + Infra) ### Plugin Registry: v2 submission path is currently unstable The community hit a real blocker: the **v2 plugin registry infrastructure** appears to be returning **404s** for both: - `elizaos-plugins/registry` (repo access), and - `plugins.elizacloud.ai` (registry endpoint) This surfaced while trying to submit `plugin-undesirables@2.0.3` (JSON metadata generated and ready). The licensing detail matters here too: **BUSL-1.1** plugins can’t be PR’d directly into the monorepo, which increases pressure on the registry path to work reliably. The team is investigating, and a policy fallback is being considered: **reverting to direct PRs into `elizaos/eliza`** if the registry flow can’t be stabilized quickly. ### Security ops: Discord account compromise + scam activity Moderators flagged **compromised admin accounts** and ongoing **scammer activity**. If you’re a community member: assume “urgent admin DMs,” “special token airdrops,” and “verification links” are hostile until proven otherwise. This week’s takeaway is simple: security work is ecosystem work. ### Agent testing in Discord: sandboxed multi-agent orchestrator request A developer requested permission to test a **Python multi-agent orchestrator** (claude-agent-sdk) in the Discord, with a constrained posture: - read-only research agent behavior - responds only to **@mentions** - no DMs - sandbox channel deployment **odilitime** offered support for **OAuth invite/whitelist** setup via DM, unblocking next steps for controlled experimentation. ### Relevant engineering context from recent repo activity (still shaping this week) While not all merges occurred this exact week, several *recently landed* platform changes are directly relevant to what the community is building and debugging now: - **Secure secrets storage is now first-class** via `@elizaos/vault` (cross-platform keychain + headless fallback), wired through Settings UI. - **Slack connector support** has been migrated into the monorepo (`@elizaos/plugin-slack`), improving cohesion across connector plugins. - **Cloud app monetization/domain plumbing** has been under active development (app-scoped chat endpoints, managed domains, container lifecycle reconciliation). If you’re building on Cloud or auto.fun, this is the substrate those experiences depend on. --- ## 3) Community Spotlight (Discord) ### Big questions on v3: roadmap clarity requested **mahmoudamer7645** raised pointed, high-value questions that many builders share: - Are we focused on **infrastructure stabilization** first, or nearing public exposure of the **full autonomous agent workflow stack**? - How much of the **identity / AgentID** system is already integrated into the upcoming architecture? These were **unanswered in the captured logs**, but they’re exactly the right questions to keep surfaced. A clear “where we are / what’s next / what’s blocked” update would help builders plan integrations and avoid churn. ### GODL + skill.md: a clean integration conversation **blankey1717** brought a concrete integration proposal (SDK + websocket streams + automation hooks + `skill.md`). **odilitime** confirmed compatibility with elizaOS’s tool/skill approach—this is the kind of pragmatic back-and-forth that turns “cool idea” into “prototype in a week.” ### Welcoming new builders + production-minded intros The introductions channel saw new members joining with full-stack and AI backgrounds, emphasizing clean code, maintainability, UX, and security. If you’re looking for collaborators, this is a good week to scroll intros and start conversations. --- ## 4) Token Economics (AI16z + auto.fun) ### auto.fun: builders’ momentum (AI Sprint results) The most tangible ecosystem signal this week is **auto.fun’s sprint conclusion** and public winners: - **1st: Build4** - **2nd: Clawdyland** - (3rd place also referenced) Even without deep technical postmortems yet, this format is doing something important: it provides a repeated loop of *ship → demo → vote → iterate*, which is exactly what an agent ecosystem needs. ### AI16z token: no official parameter changes captured this week There were **no confirmed, source-linked updates** in the provided logs about: - emissions, burns, staking mechanics, treasury actions, or - formal AI16z token economic policy changes. There *was* a community concern raised earlier about token support/signaling (e.g., social profile changes), but there’s **no recorded follow-up** in the data provided. If you want clarity, the best next step is to request an **official, pinned** token/roadmap update rather than relying on social signals. --- ## 5) Coming Soon (What to Watch) **1) v3 direction + identity/AgentID status update (needed).** Given the repeated questions, expect (and request) a concise public snapshot: what’s stabilized, what’s still in flux, and when builders can start targeting the v3 workflow stack. **2) Plugin submission policy decision.** If the registry remains inaccessible (404s), the community may see a temporary or permanent shift back to **direct PR-based plugin submissions**—especially important for third-party plugin velocity. **3) Eliza-1 0.8B training completion + deployment guidance.** The 0.8B-on-H200 training note suggests we may soon get a new “sweet spot” model for lightweight agents. What the community will want next: evals, recommended quantizations, and target devices. **4) GODL integration pilot.** If the DM thread turns into a prototype, watch for: - a reference agent that can consume GODL’s `skill.md`, - websocket event handling patterns, - and safety guidance for agent-managed wallets. **5) Moderation + security hardening.** Expect ongoing operational fixes (permissions, compromised-account recovery, scam pattern mitigation). Community trust is a platform feature. --- ## 6) Resources (Links & Pointers) ### Discord discussions (primary) - V3 roadmap questions + Eliza-1 model lineup + auto.fun sprint results (2026-05-13): https://discord.com/channels/1253563208833433701/1253563209462448241 - GODL integration proposal in **#coders** (2026-05-13): https://discord.com/channels/1253563208833433701/1300025221834739744 ### Key PRs / technical references (recently merged, actively relevant) - `@elizaos/vault` (cross-platform secrets + Settings UI): https://github.com/elizaos/eliza/pull/7197 - Slack plugin migrated into monorepo: https://github.com/elizaos/eliza/pull/7375 - Cloud monetized container app domains support: https://github.com/elizaos/eliza/pull/7376 ### Open action items the community can help with - Confirm/restore plugin registry access (repo + endpoint), and publish the definitive submission flow for BUSL-1.1 plugins. - Provide an official v3 roadmap note addressing workflow-stack exposure and AgentID integration status. - Continue coordinated Discord security response (mods + community reporting).