# ElizaOS Weekly Newsletter (2026-02-05) **Week of 2026-01-29 to 2026-02-04** ## 1) Executive Summary **1. OAuth integrations landed (and more are imminent).** This week, the team made major progress wiring OAuth into the Eliza app chat experience—**X.com, GitHub, Slack, and Linear** are now integrated, with **Notion next**. The underlying architecture is proving flexible; the main friction has been the real-world grind of credentials + redirect URI setup per provider. **2. “Autonomy by default” and skills standardization are getting clearer.** A recurring theme across Discord was that ElizaOS already has many of the capabilities people praise elsewhere (cron jobs, scheduled tasks, autonomous loops), but adoption lags. The current direction emphasizes **making autonomous mode the default** (`new AgentRuntime({ autonomous: true })`) and aligning skills with the **agentskills.io open standard** (Moltbot skills = Claude skills standard). **3. Product focus sharpened: ship revenue-driving experiences.** Leadership messaging became more explicit: build things people will pay for, and do it quickly. That urgency is shaping priorities across the framework refresh (Eliza 2.0), packaging/presentation, and new product lines like the **automated AI news/media pipeline**. --- ## 2) Development Updates ### OAuth: providers shipping, MCP next Sam reported OAuth is working for: - **X.com** - **GitHub** - **Slack** - **Linear** Next steps: - **Notion OAuth** is next on the roadmap. - **MCP testing/integration** is planned immediately after OAuth coverage is solid across major clients/adapters. A useful meta-lesson from the thread: once the OAuth scaffolding is right, adding providers is “easy”—but **provider operational details** (credentials, app approvals, redirect URIs) are the time sink. ### Skills & autonomy: closing the “feature adoption gap” Three technical threads converged: - **Autonomous mode is now “built-in,” not a plugin.** The community emphasized that the framework’s autonomy features are mature—but underused. - **Standards clarity:** Moltbot skills were confirmed to follow the same open standard as Claude skills via **agentskills.io**. - **Dynamic providers vs `skill.md`:** confirmed “functionally the same,” with providers being more strongly typed. A separate but related reliability concern emerged earlier in the week: skill invocation simply isn’t consistent enough out of the box. R0am shared a practical workaround after finding **skills failed to trigger in ~56% of eval cases**—a `UserPromptSubmit` hook that forces a 3-step “evaluate → activate Skill() → then implement” routine. It may feel clunky conversationally, but it’s a concrete path to higher task completion. ### Security and plugin ecosystem concerns Plugin momentum continues, but so do security questions. Discussions raised concerns about **malicious skills** appearing in third-party hubs. Proposed mitigations included: - “Scanner skills” to inspect packages - A code rewrite/adaptation phase before running skills in Eliza - LLM-based skill review - Potential sandboxing for skill execution ### Babylon.market internal launch: testing found real UX blockers Babylon.market moved into internal launch mode with mandatory team testing (2–3 hours + screenshots/notes). Early issues reported: - **Farcaster login failures** - **Infinite loading spinners**, especially on network switching - Waitlist position pages failing to load reliably --- ## 3) Community Spotlight (Discord) **R0am** - Brought hard numbers to the skill reliability problem and shared a working enforcement pattern that others can implement quickly. **Odilitime** - Provided clear, repeated support on token migration/bridging questions and helped users unstake on daos.fun despite confusing UI errors. - Also pushed practical dev workflow guidance (e.g., recommending certain branches/approaches) and kept conversations grounded in shippable outcomes. **Jin** - Showed the most “product-shaped” demo of the week: a **fully automated AI news show pipeline** that ingests GitHub + Discord, summarizes weekly themes, voices them through characters, and publishes episodes. The broader pitch: “agentic DAO infrastructure with a media layer” to fight information overload. **Stan ⚡** - Clarified key implementation details (providers vs skill.md), highlighted patterns already in use in Eliza Cloud, and stayed focused on standardization and integration paths (notably cskills). Notable community discussions: - A frank comparison of **ElizaOS vs OpenClaw**: OpenClaw praised for UX and self-modifying loops, but criticized for token burn and limited complexity; Eliza positioned as more business-oriented and battle-tested—just packaged poorly. - Multiple users raised **migration confusion** (notification overload), pushing documentation/communications to be treated as a first-class feature. --- ## 4) Token Economics (AI16z token + auto.fun) ### AI16Z → ElizaOS migration status - The migration window was a **hard 3-month deadline** (now closed). - Users who missed it can still **hold AI16Z or swap at their leisure**; purchases **after the snapshot won’t be auto-migrated**. - **Bridging is optional**: tokens can remain on **Solana**. ### Babylon.market + holder airdrop - Community confirmed an **airdrop to ElizaOS holders** tied to Babylon. - Waitlist dynamics (top 100 rankings shifting) sparked questions—expect more clarity once the onboarding and waitlist UX stabilizes. ### auto.fun No concrete, confirmed development updates were surfaced in the provided discussions this week beyond ongoing ecosystem coordination. If you’re building on/around auto.fun, now’s a good time to share what you need most (docs, endpoints, listing process, integrations) so priorities can be pulled forward. --- ## 5) Coming Soon - **Notion OAuth**, followed by **MCP testing** once OAuth coverage is “done enough.” - **Eliza 2.0 refresh**, with **native agent skills** and improved framework presentation. - **cskill plugin work** aimed at absorbing best-of OpenClaw-style capabilities into the Eliza skill ecosystem. - **AI media expansion** beyond the news show: talk/panel formats, Clank Tank iterations, live streams, and eventually **voice agents** for hybrid human+AI co-hosting. - **Babylon.market bug fixes**: Farcaster login, loading spinners, and waitlist status reliability. - Potential **visual consistency** solutions for image generation (LoRA or “agent-built apps” approach) to support storytelling/branding workflows. --- ## 6) Resources ### Key Discord threads - Token migration & bridging clarification: https://discord.com/channels/1253563208833433701/1301363808421543988 - Core dev discussion (OAuth, providers vs skill.md, Babylon launch issues): https://discord.com/channels/1253563208833433701/1377726087789940836 - Skills/autonomy + Moltbot standards + AI news pipeline discussion: https://discord.com/channels/1253563208833433701/1300025221834739744 https://discord.com/channels/1253563208833433701/1253563209462448241 ### GitHub items referenced this week - **elizaos/eliza** issues opened/active in Feb: - #6447 — *Eliza Character File & Prompt Engineering* - #6448 — *Billing* - PRs mentioned for review: - elizaos/eliza **PR #6457** (submitted by 0xbbjoker) - elizaOS/eliza-cloud-v2 **PR #278** (submitted by Stan ⚡) ### Practical takeaways to try now - If your agent “forgets” to use skills: implement a forced **evaluate → activate Skill() → implement** hook pattern (from R0am’s findings). - If you’re publishing skills/plugins: start thinking about **scanning + sandboxing** as table stakes, not “nice-to-have.”