# ElizaOS Weekly Newsletter (2026-02-25) *Week of **2026-02-18** to **2026-02-24*** ## 1) Executive Summary **1) A clear shift toward revenue-first shipping.** Core dev discussions outlined an org restructuring driven by tight budgets: smaller “venture-style” teams, fewer process rituals, and a focus on launching products that generate fees quickly (Milady → Eliza Cloud, Babylon trading, Hyperscape duel betting, OTC agent). **2) V2.0.0 database + isolation work accelerated.** The team is actively de-risking the **v2.0.0** upgrade path via significant database refactors—most notably **PR #6521**, which removes legacy auto-migration code and aligns isolation/RLS patterns with the established v1 approach. **3) Dev tooling and ecosystem surface area expanded again (trading + plugins).** A new agentic trading platform (**fomolt**) was introduced for testing Eliza agents on Base via a CLI (paymaster + API keys), and a new exchange BD plugin (**SolCex**) was published and queued for registry inclusion. --- ## 2) Development Updates (Core + Plugins) ### V2.0.0 migration strategy: less magic, more reliability - **DB refactor / migration simplification (PR #6521)**: ~2,600 lines removed, largely by dropping auto-migrations from **1.4.x → 1.6.x**. The assumption is most users have already moved past 1.4.x; a future improvement may be to **hard-fail with a clear error** telling users to migrate to 1.6.x before attempting v2. - **RLS (Row Level Security) remains in scope**: Odilitime confirmed their parallel “great db refactor” work preserves RLS expectations while staying easy to rebase against Stan’s changes. ### Performance + prompt-bloat pressure: token limits are a real pain - Odilitime reported repeatedly hitting **~200k token limits** when many plugins are enabled—pinpointing **bootstrap providers and evaluations** as frequent culprits. - Related work in the repo this month includes efforts that reduce prompt bloat and runtime errors: - **Action filtering by relevance (PR #6475)**: introduces an `ActionFilterService` (vector search + BM25 reranking) to shrink “200+ actions” down to a small relevant set (~15), which should directly help context size. - **Bootstrap robustness and caching improvements (PR #6476)** plus defensive fixes like: - **Null checks in `@elizaos/core` settings utilities (PR #6471)** - **Null check for `runtime.providers` in plugin-bootstrap (PR #6473)** ### New plugin + Web3 capabilities - **SolCex Exchange BD plugin published**: `@buzzbd/plugin-solcex-bd` (npm), an “autonomous business development agent” for exchange listings: - Integrates **8 services**, uses a **100-point token scoring** model, and pulls from **16 intel sources** - Registry submission: **elizaos-plugins/registry PR #263** - **NFT capability discussions**: current limitations acknowledged, with ongoing **plugin-evm** work expected to unlock broader NFT actions later (and some community interest in a potential “Eliza NFT” concept). ### Tooling: “fomolt” agentic trading testbed on Base Fido introduced **fomolt**, focused on fast iteration for agent trading workflows: - CLI for instant trades on **Base** - **USDC for gas** via paymaster support - **API-key auth** (no seed phrase handling in dev loops) Community feedback was explicitly requested—this is a good moment for builders to shape the interface before it hardens. --- ## 3) Community Spotlight (Discord) ### Reassurance during a rough market Price concerns surfaced in public channels (“is the project dead?”). **Odilitime** responded directly: development is active; price weakness is largely attributed to broader bear-market conditions. This was an important morale stabilizer, even as members noted the ongoing “price vs. activity” disconnect. ### Builders helping builders - **Stan ⚡ + Odilitime** collaborated on optimization thinking around MCP loading and METATOOL search/action patterns. The takeaway: token blow-ups may be more about *recent messages + reflections collecting too much data* than actions alone. - **DarmaStef** showcased real-world agent builds (reservation, travel, database analysis for non-technical users) and offered to collaborate—great proof that ElizaOS can support practical assistants, not just demos. ### Practical ops: roles, permissions, and scam hygiene - **Odilitime** unblocked **shad0w**’s Discord access (xfn + milady rooms) after role assignment limitations required manual channel permission fixes—small work, but it keeps collaboration moving. - Community members reiterated **anti-scam practices**: avoid DM links and fake “ticket” systems during any token or migration-related confusion. --- ## 4) Token Economics (AI16z token + auto.fun) **Market sentiment:** Community conversation this week was dominated less by protocol mechanics and more by **price performance concerns** and long-term confidence. Team messaging emphasized that shipping continues despite bearish conditions. **Token-holder “upgrade” teaser:** Odilitime shared news of a token-holder upgrade referenced from a tweet (via shawmakesmagic), but the community still had open questions about *which asset* the upgrade applies to (ElizaOS vs. $milady.ai). Treat this as **unconfirmed until clarified in official posts**. **Milady token status:** A key clarification: **the `milady.ai` token has not launched yet**, which explains why users couldn’t find it in wallets like Phantom. **auto.fun:** No concrete, verifiable development notes landed in the provided logs this week. If you’re tracking auto.fun closely, the best move is to monitor official announcements and verified links only (and keep skepticism high around anything circulating via DMs). --- ## 5) Coming Soon ### Product launches and integrations (revenue-first roadmap) Based on core-dev planning, expect near-term movement on: - **Agent Arena duel betting in Hyperscape** (target: next week; fee model discussed as **2% + 1%**) - **Babylon trading platform rollout** ramping from **1% to 100%** user exposure (with **1% fees** referenced) - **OTC agent at `tradingdesk.fun`** (referenced **1% fee** model) - **Milady shipping** as a funnel to **Eliza Cloud** (positioned as the primary revenue driver) - **Eliza App** onboarding + billing completion so charging can begin - “**Jeju**”: a network for agents (post-Milady focus) ### Developer experience improvements to watch - Continued work to reduce context blow-ups (providers/evaluations), plus potential future “metatool/provider” patterns if the ROI makes sense. - More guidance likely around **safe migrations** to v2.0.0, especially with legacy auto-migrate paths being removed. --- ## 6) Resources (Links) ### Discord threads (high-signal) - Coders channel (fomolt + ops): https://discord.com/channels/1253563208833433701/1300025221834739744 - Discussion channel (project status + PumpFun mention): https://discord.com/channels/1253563208833433701/1253563209462448241 ### Key PRs / workstreams (GitHub) - DB refactor / migration removal (v2.0.0): **elizaos/eliza PR #6521** - Reduce action prompt bloat: **elizaos/eliza PR #6475** - Bootstrap optimizations: **elizaos/eliza PR #6476** - Defensive runtime fixes: **elizaos/eliza PR #6471**, **#6473** - SolCex plugin registry submission: **elizaos-plugins/registry PR #263** ### Open issues worth awareness (cost + UX) - Duplicate LLM calls when messages include URLs: **elizaos/eliza Issue #6486** - Custom OpenAI-compatible endpoint support request: **elizaos/eliza Issue #6490** --- *That’s it for this week. If you’re building: try **fomolt** with an Eliza agent, keep an eye on the **v2 DB migration** changes, and consider testing with many plugins enabled to help reproduce and squash the token-limit pain points.*