# ElizaOS Weekly Newsletter (2026-03-12 to 2026-03-18) ## 1) Executive Summary **1. Plugin ecosystem momentum is accelerating.** This week featured multiple concrete pushes toward a richer, more composable ElizaOS ecosystem: expanded EVM/on-chain data access work (via `plugin-evm` + goldrush.dev), plus a promising new “agent browser” concept (**unbrowse**) that could become a high-impact plugin for web-scale agent retrieval. **2. v2.0.0 architecture conversations are getting sharper (and healthier).** The community discussed how to avoid the “uncontrolled submissions” problem from the 0.x era by potentially **shipping v2.0.0 with zero default skills**, and instead enabling decentralized skill discovery/hosting (e.g., `yourdomains.com/skills.md`). This is an important decision about keeping core clean while still enabling rapid community innovation. **3. Token + product strategy took center stage—along with requests for transparency.** There was strong community concern about token performance and marketing focus. In response, Odilitime clarified the intended business flywheel: **Milady distribution → elizacloud revenue → ElizaOS flywheel**, with an **elizacloud launch targeted for end of week / early next week**. In parallel, the community raised unresolved questions about **ai16z → ElizaOS migration transparency** and **which Milady chain (SOL vs BSC) is officially supported**. --- ## 2) Development Updates (Framework, Plugins, Fixes) ### Plugin development & infrastructure - **`plugin-evm` progress:** Community member **dinesh** shared active work to: - Add **100+ on-chain data support** via **goldrush.dev** integration - Fix open issues in the `plugin-evm` component Net effect: easier access to richer on-chain context for agents (balances, transactions, protocol signals, etc.), which strengthens ElizaOS as a “real-world connected” agent framework rather than a purely chat-centric one. ### v2.0.0 “skills” structure and governance - **Skills folder work (PR mentioned: #6597):** Odilitime raised a critical release-quality question: how to prevent skill submissions from turning into bloat or low-signal noise, as happened with earlier plugin systems. - **Proposed direction:** ship **v2.0.0 with 0 default skills** and promote a decentralized discovery pattern (community-hosted indices like `skills.md`), rather than centralizing everything in core. If adopted, this could keep the core repo stable and auditable while allowing experimentation to flourish at the edges—an approach that tends to scale better as the community grows. ### Integration proposals worth watching - **Effect AI (human-in-the-loop task marketplace):** Miguel from Effect AI proposed an integration where ElizaOS agents can **post tasks to a decentralized human worker network** when automation hits limits (labeling, review, translation, annotation, etc.). This is a meaningful “agent reliability” lever: instead of failing or hallucinating when uncertain, agents can route to verified human workstreams. - **Z1N Protocol (on-chain signaling / identity persistence):** Z1N introduced an on-chain signaling concept for “NBI participation,” with **soulbound Keys** and epoch-based attestations. Even if the exact mechanism evolves, the underlying problem—**agent identity continuity across sessions**—is a real need as agents become more autonomous and accountable. ### Bug/ops: onboarding friction - **Role assignment Google Form failure:** Users reported an **“Invalid Dynamic Link”** error in the role assignment flow, consistent with a misconfigured Firebase Dynamic Links domain or malformed URI handling. This is small but important: onboarding friction directly reduces community conversion and contributor throughput. --- ## 3) Community Spotlight (Notable contributions & Discord discussions) ### Content & communication - **Jin’s weekly roundup trailer** landed well and got positive community feedback—exactly the kind of lightweight, repeatable media that helps keep builders aligned on “what shipped” and “what’s next.” ### “unbrowse” — a fast browser for agents - **lekt9** introduced **unbrowse**, described as a browser that runs at ~**100× speed** by traversing APIs rather than DOM, while passively indexing web APIs for agent use. - **Odilitime’s guidance** was pragmatic and actionable: integrate via the **ElizaOS plugin registry** and follow the plugin creation docs; Cursor can help generate scaffolding quickly. Takeaway: this is a strong example of “community R&D → plugin distribution path,” which is exactly the loop we want. ### Production readiness: from demo magic to enterprise trust - A thread kicked off by **Caesar ⚔️** emphasized that the bottleneck is increasingly **trust + polish**, not raw capability: - UI trust signals that feel “enterprise” - Robust error handling and recoveries - Context persistence across sessions - Localization concerns (including Arabic RTL) This aligns with what many teams are learning: users don’t adopt autonomy until systems *feel* dependable. ### Security for autonomous DeFi agents - **x402Guard** was discussed as a guardrail layer for DeFi agents: - **Per-step evaluation** (not whole-strategy pre-validation) - Layered limits: permanent guardrails + temporary session keys - Immutable audit log of *every* attempt This is exactly the kind of practical “agent safety engineering” that can differentiate ElizaOS-aligned tooling. --- ## 4) Token Economics (AI16z token + auto.fun + ecosystem economics) ### ELIZA token sentiment & strategy clarification - Community sentiment was candid: frustration that ELIZA price action has lagged while broader AI-crypto markets move. Critiques focused on marketing clarity and perceived attention dilution. - **Odilitime clarified the intended flywheel:** Shaw promotes **Milady** → drives distribution for **elizacloud** → elizacloud revenue supports the **ElizaOS** ecosystem. A launch is targeted **end of week / early next week**. ### Open token migration questions (ai16z → ElizaOS) - A community member raised unresolved transparency concerns: - Estimated migration may be low (5–10%) with a large number of addresses still holding ai16z - Questions on the actual migration percentage, the post-migration token breakdown, and plans for unmigrated allocations These questions remained unanswered in the captured discussions and are important to resolve clearly. ### auto.fun - No concrete, verifiable updates on **auto.fun** were captured in the provided logs this week. If there are new releases, metrics, or roadmap changes, share them in Discord so they can be included next issue. --- ## 5) Coming Soon (What to anticipate next) - **elizacloud launch** (target: end of week / early next week, per Odilitime) - **unbrowse as an ElizaOS plugin** (high potential: faster web/API traversal + API indexing for agents) - A decision on **v2.0.0 skills policy** (ship with zero skills vs curated defaults; and how decentralized `skills.md` discovery should work) - Possible start of a **Clawhub-style directory/registry** that includes both skills and plugins - Fixes to **role assignment onboarding** (Google Form Dynamic Link configuration) - Clarity items the community is explicitly asking for: - **Which Milady chain is officially supported (SOL or BSC)?** - **ai16z migration transparency** and next steps --- ## 6) Resources (Links & references) - **Plugin Registry (distribution + packaging):** https://github.com/elizaos-plugins/registry - **Plugin documentation:** https://docs.elizaos.ai - **Jin’s weekly roundup trailer (media):** https://cdn.elizaos.news/elizaos-media/2026-03-15_trailer_2106a81a.mp4 - **Discussion reference (Discord channel):** https://discord.com/channels/1253563208833433701/1253563209462448241 If you’re building this week: consider picking one “ecosystem brick” (a plugin, a skill, or a directory entry) and making it shippable via the registry path. Small, composable contributions are adding up fast.