# ElizaOS Weekly Newsletter (Apr 5 – Apr 11, 2026)

## 1) Executive Summary

This week pushed ElizaOS further into “agents with real-world power” territory—more finance, more interoperability, and a sharper focus on *trust + safety*.

- **Cross-chain & agent finance accelerated.** Work continued on agent-native financial tooling, including progress toward cross-chain swap coverage via new registry submissions (notably MangoSwap) and the broader push for non-custodial, multi-network agent operations.
- **Trust/identity became the center of gravity.** The community and contributors leaned into the “cold-start trust” problem (how unknown agents can verify one another) with proposals around attestations, on-chain credentials, and watermarking—aligning with the project’s strategic shift toward cryptographic identity (AgentID) over pure messaging.
- **Eliza v3 (v2.x) moved closer to release.** Core devs confirmed new v3 agents are actively being tested, with code already available in the `develop` branch and integrated into Milady—release/announcement pending test completion.

## 2) Development Updates (Core + Plugins)

### Core runtime & agent experience
- **TOON action parameters + fewer “spammy” continuations (PR open):**  
  PR **#6709** fixes a real usability bug for connectors using TOON encapsulation (e.g., Discord/Milady): required action params were not being requested in the schema, causing tool calls like `RUN_IN_TERMINAL` to miss critical inputs. It also marks several async “task spawning” actions as terminal to prevent the continuation loop from generating filler responses while background PTY tasks run.
- **New local “agent harness” workflow (PR open):**  
  PR **#6702** introduces an `agent/` workspace REPL harness for booting the repo locally, plus runtime composition improvements (loading characters from JSON file paths, adding `checkShouldRespond`, and tooling for local plugin submodules). Review notes flagged install/workspace consistency issues that should be resolved before merge (submodule workspace paths and lockfile mismatches).

### Plugin ecosystem: interoperability, safety, and economics
- **Cross-chain swaps expanded via registry work:**  
  A MangoSwap plugin submission was initiated in the plugins registry to enable cross-chain swaps across **10+ networks** (a meaningful step toward “agents that can route value anywhere”).
- **Anthropic plugin compatibility work:**  
  Ongoing changes target **Opus 4.x** compatibility—temperature controls, output caps, and updated naming aligned with `ai-sdk v6`.
- **Windows developer experience improvements:**  
  `plugin-openrouter` work continued to remove artifacts that were blocking Windows git checkouts—important for smoother onboarding across platforms.

### Security & dependency hygiene
- **Security-minded dependency upgrades (PR open):**  
  Dependabot PR **#6694** bumps `path-to-regexp` and other deps, addressing multiple CVEs and tightening route-pattern handling (worth prioritizing given how broadly these libraries can surface in web tooling).

### Proposals and WIP contributions to watch
- **Economic memory plugin (PR open, needs hardening):**  
  PR **#6701** proposes `plugin-mnemopay` to give agents “economic memory” (learning from settlements/refunds, building reputation). Automated review raised important concerns (persistence, NaN config hazards, null-safety, unbounded growth, no tests). Strong concept—needs engineering rigor before it becomes production-grade.
- **Safety checks for trading (issue/proposal):**  
  Issue **#6706** proposes `plugin-safeagent` for multi-chain token safety checks and pre-trade blocking—directly relevant to Discord’s repeated “agent safety” concerns.
- **Capability enforcement / authorization layer (issue/proposal):**  
  Issue **#6707** proposes cryptographic capability tokens and tiered approvals for tool calls—highly aligned with the community’s open question: *“How do we stop agents from doing unsafe things?”*

## 3) Community Spotlight (Discord Highlights)

### Shipping wins & onboarding moments
- **First agent deployed (Hatcher.host):**  
  `jgonly1_89829` reported a successful one-click deploy with Groq integration, clean dashboard, and live logs—exactly the kind of “time-to-first-agent” success story we want more of.
- **New builder intro:**  
  `0xcrackedev` (fullstack dev, ~5 years in AI products + OSS) introduced themselves and shared profiles—welcome! If you’re looking for tasks, the open PR/issue list below is a great starting point.

### Product clarity: multichain, not a migration narrative
A recurring confusion popped up—“Is ElizaOS moving from Solana to ETH?” Community members clarified: **ElizaOS is multichain**, not exclusive to one chain. That framing matters as more plugins and wallets aim to route assets across ecosystems.

### Important open questions: safety, trust, and inter-agent coordination
- In `#coders`, a key question remained unresolved: **how to prevent agents from performing unsafe operations** as capabilities expand (transactions, tool calls, automation). This dovetails with the week’s broader AgentID + trust discussions and the new capability-token proposals.
- Separately, architectural discussion highlighted the **cold-start trust problem**: how independent agents can establish verifiable trust without shared history. Proposed directions included **WTRMRK**, **on-chain credentials**, and **append-only attestation sequences** (notably on Base L2) to build an audit-able behavioral timeline.

### Market sentiment & moderation vigilance
There was a heavy dose of token-price speculation and negative sentiment (including comparisons between ELISA and Ai16z). Also: several users were flagged as scammers—thanks to those staying vigilant and reporting.

## 4) Token Economics (AI16z, auto.fun, and adjacent initiatives)

No official tokenomics changes were announced this week for **AI16z** or **auto.fun**, but a few ecosystem signals are worth noting:

- **Community sentiment is volatile.** Discord saw intense discussion around token drawdowns and broader market fatigue. If you’re contributing, focus on fundamentals: shipping, usability, safety, and integrations.
- **Incentive and “agent economy” proposals are increasing.** The **AIGEN Protocol** proposal (**#6708**) suggests token incentives for contributing agent tools and safety checks. Regardless of whether AIGEN becomes “official,” it reflects a growing desire to tie *useful agent work* to *economic rewards*—a theme consistent with ElizaOS’s shift toward an economic layer for agents.
- **auto.fun remains part of community identity.** You’ll still see “auto.fun enjoyer” energy in roles and conversation, but this week’s concrete updates were more on the engineering side than product announcements.

## 5) Coming Soon (What to anticipate)

- **Eliza v3 (v2.x) release + announcement** once new agents finish testing (code already in `develop`, integrated into Milady).
- **More cross-chain “agent finance” primitives** as swap plugins mature and wallet tooling expands across networks.
- **A clearer safety story**: expect continued debate (and hopefully implementation) around capability-based authorization, guardrails, and audit trails—especially for on-chain actions.
- **ai-news pipeline improvements**: planned fallback models and better credit-usage monitoring for data regeneration.

## 6) Resources (Links worth your time)

### Key GitHub PRs
- TOON params + async continuation fixes: https://github.com/elizaos/eliza/pull/6709  
- Local agent harness (`agent/`) + runtime composition: https://github.com/elizaos/eliza/pull/6702  
- MnemoPay economic memory plugin (WIP): https://github.com/elizaos/eliza/pull/6701  
- Dependency security bumps (CVE-related): https://github.com/elizaos/eliza/pull/6694  

### Notable GitHub issues / proposals
- AgentID (trust & identity direction): https://github.com/elizaos/eliza/issues/6688  
- Capability token enforcement (authorization layer): https://github.com/elizaos/eliza/issues/6707  
- SafeAgent token safety checks: https://github.com/elizaos/eliza/issues/6706  
- `elizaos create` macOS/Bun install failure report: https://github.com/elizaos/eliza/issues/6704  
- AIGEN Protocol incentive proposal: https://github.com/elizaos/eliza/issues/6708  

### Discord threads (context & discussion)
- General discussion (multichain clarification, onboarding, sentiment): https://discord.com/channels/1253563208833433701/1253563209462448241  
- Coders channel (v3 testing, safety question, trust architecture): https://discord.com/channels/1253563208833433701/1300025221834739744