# ElizaOS Weekly Newsletter (2026-02-21) **Week of 2026-02-15 → 2026-02-21** ## 1) Executive Summary **1) Beta-launch readiness accelerated (core stability + onboarding).** This week’s engineering focus continued the “clear blockers, ship fundamentals” push: documentation expanded, reliability fixes landed across core and plugins, and the project kept tightening the path from “install” to “running agent” for upcoming beta testers. **2) Agents gained reach and identity: more connectors + on-chain agent passports.** Key integrations (WhatsApp, Gmail, n8n) remain central to the “agents where people already work” strategy, and the **SAID Protocol** initiative is setting up verifiable, discoverable identities for agents on Solana—an important building block for reputation, coordination, and on-chain actions. **3) Babylon momentum + community visibility improved.** Babylon opened access to the **top 5,000 waitlist users**, and ElizaOS also received a **golden checkmark verification** (shared by ElizaBAO). Both helped address community concerns about delivery after recent Autofun/Twitter turbulence. --- ## 2) Development Updates (Technical) ### Core framework + platform direction - **V2 / “next” track continues to mature**: multi-language foundations (Rust, Python, TypeScript) and a leaner runtime-first posture are taking shape in large ongoing PRs (not yet merged). This work aims to reduce “everything bundled” complexity and make ElizaOS easier to embed into other stacks. - **SAID Protocol (agent identity on Solana)**: work progressed on automatically creating an agent identity during `elizaos create` (PR draft). The idea: give every agent a “passport” (keypair + registry entry) with an opt-out design so it won’t break local setups. - **Action selection got smarter**: the new **ActionFilterService** (vector search + BM25 reranking) reduces prompt bloat by narrowing the action set shown to the model (helpful once agents have hundreds of actions/providers installed). ### Reliability, typing, and maintenance - **Bootstrap plugin robustness/performance**: continued optimization work (including caching/robustness improvements) to keep runtimes stable as plugin counts grow. - **Strict typing refactor in logger runtime**: a dedicated refactor PR focuses on replacing loose `any`-style typing with stronger type safety—small change in appearance, big change for long-term maintainability. - **Dependency updates across Rust + Node projects**: routine bumps landed/queued to keep builds secure and reproducible. ### Workflow automation: n8n plugin stability - A **critical fix** addressed “multi-step confabulation” problems in the n8n workflow plugin—important for anyone chaining tool calls where partial/incorrect state can create cascading failures. ### Product/UI notes from daily dev status - **Landing page**: Discord was added as a **fourth messaging option** on the main site. - **Metrics**: baseline product metrics were established to track beta launch performance. - **New widget work**: a chatbot widget in progress (“**Vera — Virtually Ever After**”) suggests continued investment in embeddable agent experiences. --- ## 3) Community Spotlight (Discord) ### High-signal bug reporting: ElizaCloud dashboard Community member **DorianD** did hands-on testing and surfaced several **UI/UX and API explorer issues** that are now in the team’s queue: - Broken mousepad scrolling on the API explorer (scrollbar is hard to see): https://www.elizacloud.ai/dashboard/api-explorer - “**Send request**” test tool returns **“api key is required”** even when a key appears present - Requests for **per-model price breakdowns** in-dashboard - Questions about whether **“use a different key”** means true **BYOK** (Bring Your Own Key) like OpenRouter **Odilitime** acknowledged the report and forwarded it to the dev team—exactly the kind of tight feedback loop we want heading into beta. ### Security builders: early rugpull detection wins **Vlt9** shared results from live testing of a **Solana rugpull detection system** that reportedly flagged two major incidents *~2 hours before crashes* (DARKTRUMP marked High-Risk at peak; also NVIDIA token). They added documentation and offered test keys to collaborators—if you’re building risk tooling, this is one to watch. ### Ecosystem + docs tooling discussion - **Plugin development guidance**: Kenk pointed builders to the official “create a plugin” guide and reinforced the norm: if it’s a reusable service integration, **ship it as a plugin** rather than waiting for a ticketing process. - **Documentation workflow idea**: Stan proposed using **HackMD MCP** to reduce copy/paste and improve shared team notes—an example of the community pushing on “dev velocity” beyond code. ### Product access and event presence - **Babylon** opened to the **top 5,000** on the waitlist. - EthDenver presence: Shaw was confirmed to be participating in **ETHBuilder talks**, with coordination offers from community members. --- ## 4) Token Economics (AI16z / ElizaOS token + auto.fun) This week’s token conversation was candid—and useful. ### Key clarifications (and remaining confusion) - Community members confirmed that **$ai6z / ai16z and $elizaos are different contracts**, and that ai16z is worth significantly less (per Odilitime). Several people requested the ElizaOS contract address and were directed to the appropriate channel. - Migration questions persist: - “Can I still migrate if I missed the deadline?” - “How many users still need to migrate?” - “Why is the token mintable?” (unanswered in-channel this week) ### Utility + narrative feedback (from community) - Multiple members argued that the rebrand cost some “memetic” momentum and that token utility is still unclear to outsiders. - The discontinued “tribute” model (new agents contributing 5–10% to the DAO) was discussed as impractical given liquidity/usage constraints. - Broader sentiment: **2026 is a building year**, and recovery likely depends on **ElizaCloud user growth** and the **Jeju network** becoming operational—plus more “best-in-class, well-documented APIs” that developers can trust. ### auto.fun context While there weren’t major new auto.fun release notes in the logs provided, the **Autofun incident** remains a reference point in community trust and delivery expectations. The clearest actionable takeaway: product wins (like Babylon access) need to be paired with tighter, more centralized communication. --- ## 5) Coming Soon (What to anticipate) - **ElizaCloud UX + API explorer fixes**: scrolling and API test tooling are high priority given they block developer self-serve. - **Pricing transparency + BYOK clarification**: expect follow-ups clarifying “use a different key,” plus clearer per-model pricing to reduce onboarding friction. - **Beta onboarding polish**: documentation, metrics, and stability work suggests the team is preparing for more structured testing. - **Multi-agent coordination discussions**: community threads on cross-instance communication patterns hint at upcoming architectural guidance for multi-agent systems. - **Improved partner communications**: the 🥇-partners channel called out communication gaps causing avoidable negative sentiment—expect process improvements here. --- ## 6) Resources (Links) **ElizaCloud** - API Explorer bug report target: https://www.elizacloud.ai/dashboard/api-explorer **Docs** - Plugin creation guide: https://docs.elizaos.ai/guides/create-a-plugin **Selected GitHub work referenced this week (for deeper reading)** - SAID Protocol agent identity (draft): https://github.com/elizaos/eliza/pull/6510 - Action filtering via vector search + BM25: https://github.com/elizaos/eliza/pull/6475 - Bootstrap plugin optimization work: https://github.com/elizaos/eliza/pull/6476 - Environment variables documentation: https://github.com/elizaos/eliza/pull/6377 - Core documentation guides bundle: https://github.com/elizaos/eliza/pull/6356 **Discord threads to revisit** - Token clarification + migration discussion (Feb 19–20) - ElizaCloud bug report + pricing/BYOK questions (Feb 20) - Rugpull detection testing + doc drop (Feb 18)