Skip to content
agentboards
Browse sections News

Agent news & practical updates

Latest tracked release: Nov 16, 2026 · Source links included · Dates in UTC

Major & notable updates · past 30 days

Showing 5 of 24 matching updates.

Major update· Harness

Browser workbench where an orchestrator CLI agent delegates to worker CLI agents, all of them real PTY processes on your own machine

What changed. Include Kimi Code CLI support and Windows terminal scrolling fixes merged since 2.2.1.

Why it matters. Check the release's compatibility and migration changes before upgrading.

Profile context & before you switch

Where it stands out

Supports multi-agent workflows. Lets you bring your own model.

Before you switch

Source-available under BUSL-1.1, converting to Apache-2.0 on 2030-05-16; it uses the agent CLIs already installed and authenticated on your machine

Runs local · macos, linux, windows, web.

Slug is hivehq to avoid a collision with the unrelated Aden Hive agent runtime already listed on the board.

Versions at or before 0.6.0-alpha.7 were Apache-2.0; from 0.6.0-alpha.8 the project is BUSL-1.1, converting to Apache-2.0 on 2030-05-16.

Compare with GitLab Duo and Pane

Same-category alternatives, matched by execution, platform and capabilities. Open the full comparison for source dates and qualifications.

ProfileHiveGitLab DuoPane
Runslocallocal / cloudlocal / cloud
PricingSource-available under BUSL-1.1, converting to Apache-2.0 on 2030-05-16; it uses the agent CLIs already installed and authenticated on your machineDuo Core is included with Premium and Ultimate subscriptions; Duo Pro and Duo Enterprise are per-seat add-ons, and Agent Platform usage is billed with GitLab CreditsFree and open source under AGPL-3.0; you pay for the coding CLIs and model subscriptions it runs
Local modelsNot listedSupportedSupported
Open sourceNoNoAGPL-3.0
Notable update· Terminal

Go terminal coding agent with resumable sessions, multi-agent support and zero-config mDNS LAN chat between instances

What changed. Material deprecation advisory: analyze reports 16 new info-level material.dart -> material_ui deprecations on 3.50 (progressive migration path, non-blocking; CI mobile-test runs tests only).

Why it matters. Check whether the new capability fits a task you already use this agent for.

Profile context & before you switch

Where it stands out

Supports multi-agent workflows. Lets you bring your own model. Open source · MIT.

Before you switch

Free and open source under MIT; you pay the model provider you configure

Runs local · macos, linux, windows.

Compare with Codex CLI and Claude Code

Same-category alternatives, matched by execution, platform and capabilities. Open the full comparison for source dates and qualifications.

ProfileggcodeCodex CLIClaude Code
Runslocallocallocal / cloud
PricingFree and open source under MIT; you pay the model provider you configure Included with ChatGPT Free, Go ($8), Plus ($20), Pro (from $100), Business and Enterprise; or pay per token with an OpenAI API keyIncluded with Claude Pro and Max subscriptions, or pay-as-you-go through an API key
Local modelsNot listedSupportedNot listed
Open sourceMITApache-2.0No
Notable update· Harness

Open-source personal agent harness that runs on your own devices and answers on WhatsApp, Telegram, Slack, Discord and more

What changed. Safer upgrades: Prevent a newer CLI from migrating a shared database beyond an older installed Gateway daemon while allowing unrelated installation databases to remain independent. #186971, #186898, #187819, #187870, #187886, #187878, #187966, #187937, #190194, #190177 Authored

Delegated administration: Require approval for persistent delegated changes without effective authority, support explicit global or per-agent standing policy, and recheck the grant at each persistent effect. #196984 Thanks @RomneyDa and @joshavant.

Sandbox protection: Protect nested skill mounts from writable ancestors, prepare missing protected sources on first launch, and resume turns after upgrade adds protected mounts. #180840, #182412, #94425, #182789 Authored

Why it matters. Check whether the new capability fits a task you already use this agent for.

Profile context & before you switch

Where it stands out

Supports local models. Includes Docker sandboxing. Supports multi-agent workflows.

Before you switch

Free and MIT-licensed; you bring your own hosted-model keys or run local models

Runs local / sandbox · macos, linux, windows.

Compare with Hermes Agent and DeerFlow

Same-category alternatives, matched by execution, platform and capabilities. Open the full comparison for source dates and qualifications.

ProfileOpenClawHermes AgentDeerFlow
Runslocal / sandboxlocal / cloud / sandboxlocal / sandbox
PricingFree and MIT-licensed; you bring your own hosted-model keys or run local modelsFree and MIT-licensed with your own keys (OpenRouter, OpenAI, Anthropic or any endpoint); optional Nous Portal subscription bundles 300+ models and hosted tools from $20/mo, with a free tierFree and MIT-licensed; self-hosted with your own model keys and no usage fees
Local modelsSupportedSupportedSupported
Open sourceMITMITMIT
Notable update· Terminal

Coding agent built on Pi with the IDE wired in, from LSP and a debugger to a headless browser and subagent fan-out

What changed. Added publicly available image-sizing helpers for determining model-delivered image dimensions and fitting images to a specified long edge before coordinate mapping.

Restricted reasoning-effort options for Mistral GLM 5.3, GLM 5.3 Flash, and Mistral Large 4 to the levels supported by those models.

Added opt-in mnemopi.recallRelevance (off, auto, or on; default off) to filter automatic memory recall for task relevance using the judge role.

Why it matters. Relevant if your agent needs to carry information across longer tasks or sessions.

Profile context & before you switch

Where it stands out

Supports local models. Supports headless or CI workflows. Supports multi-agent workflows.

Before you switch

Free and MIT-licensed; use API keys, local models, or existing Cursor, Copilot and Devin subscriptions

Runs local · macos, linux, windows.

Compare with Claude Code and Zero

Same-category alternatives, matched by execution, platform and capabilities. Open the full comparison for source dates and qualifications.

ProfileOh My PiClaude CodeZero
Runslocallocal / cloudlocal / sandbox
PricingFree and MIT-licensed; use API keys, local models, or existing Cursor, Copilot and Devin subscriptionsIncluded with Claude Pro and Max subscriptions, or pay-as-you-go through an API keyFree and open source under MIT; bring your own provider key or run a local model
Local modelsSupportedNot listedSupported
Open sourceMITNoMIT
Notable update· Harness

Source-available desktop IDE and CLI for running 100+ CLI coding agents in parallel across isolated git worktrees

What changed. feat(plugins): add Ahrefs MCP plugin

Why it matters. Worth comparing if you connect your agent to external tools and services.

Profile context & before you switch

Where it stands out

Supports headless or CI workflows. Supports multi-agent workflows.

Before you switch

Free for one user with local workspaces, desktop app, GitHub integration and CLI; Pro $15/user/mo billed yearly or $20 monthly adds unlimited users, remote access, Linear and Slack; Enterprise custom with SSO, SCIM and audit logs

Runs local · macos, linux.

Compare with Hermes Agent and Trinity

Same-category alternatives, matched by execution, platform and capabilities. Open the full comparison for source dates and qualifications.

ProfileSupersetHermes AgentTrinity
Runslocallocal / cloud / sandboxlocal / cloud / sandbox
PricingFree for one user with local workspaces, desktop app, GitHub integration and CLI; Pro $15/user/mo billed yearly or $20 monthly adds unlimited users, remote access, Linear and Slack; Enterprise custom with SSO, SCIM and audit logsFree and MIT-licensed with your own keys (OpenRouter, OpenAI, Anthropic or any endpoint); optional Nous Portal subscription bundles 300+ models and hosted tools from $20/mo, with a free tierFree and open source under Apache-2.0; you bring an Anthropic or Google API key for the agents it runs
Local modelsNot listedSupportedNot listed
Open sourceNoMITApache-2.0

The two-minute briefing

No current editorial picks. Browse the source-linked updates below for the latest tracked changes.

New to AgentBoards · past 30 days

Recently added to the directory. A listing date is not a product launch date.

New to the board· Terminal

MIT-licensed terminal coding harness with October Bus coordination, custom providers and headless interfaces

Why it matters. New to this directory—not necessarily newly launched. Compare its workflow and requirements with tools in the same category.

Profile context & before you switch

Where it stands out

Supports local models. The custom-model guide documents Ollama, LM Studio and vLLM endpoints. The underlying runtime still determines model compatibility and resource requirements. Supports headless or CI workflows. Supports multi-agent workflows. Coordination is documented through Bus peers. Public Bus delivery is pull-based and does not automatically wake an idle agent.

Before you switch

MIT-licensed source code; use October inference or another supported provider. Provider subscriptions and inference are separate from the software license.

Runs local · macos, linux.

Free source code does not establish a lasting free inference allowance.

The custom-model guide documents Ollama, LM Studio and vLLM endpoints. The underlying runtime still determines model compatibility and resource requirements.

Compare with Codex CLI and Claude Code

Same-category alternatives, matched by execution, platform and capabilities. Open the full comparison for source dates and qualifications.

ProfileOctober HarnessCodex CLIClaude Code
Runslocallocallocal / cloud
PricingMIT-licensed source code; use October inference or another supported provider. Provider subscriptions and inference are separate from the software license.Included with ChatGPT Free, Go ($8), Plus ($20), Pro (from $100), Business and Enterprise; or pay per token with an OpenAI API keyIncluded with Claude Pro and Max subscriptions, or pay-as-you-go through an API key
Local modelsSupportedSupportedNot listed
Open sourceMITApache-2.0No
New to the board· IDE

MIT-licensed desktop workspace for existing coding CLIs, files, terminals and browser previews; beta

Why it matters. New to this directory—not necessarily newly launched. Compare its workflow and requirements with tools in the same category.

Profile context & before you switch

Where it stands out

Lets you bring your own model. Open source · MIT.

Before you switch

MIT-licensed source and free beta application. Connected coding-agent subscriptions and API usage are billed by their providers; a permanent inference allowance is not established.

Runs local · macos, windows.

Free beta software does not establish free use of a connected model provider.

Native local-model support was not established in the reviewed setup guide.

Compare with Zed and Theia IDE

Same-category alternatives, matched by execution, platform and capabilities. Open the full comparison for source dates and qualifications.

ProfileBetterC0deZedTheia IDE
Runslocallocallocal / cloud
PricingMIT-licensed source and free beta application. Connected coding-agent subscriptions and API usage are billed by their providers; a permanent inference allowance is not established.Personal plan is free with unlimited use of your own API keys or external agents; Pro $10/mo with $5 of tokens included; Business $30/seat/moThe IDE and framework are free and open source; you pay only for the LLM provider you configure
Local modelsNot listedSupportedSupported
Open sourceMITGPL-3.0-or-laterEPL-2.0
New to the board· Harness

Self-hosted local AI studio with desktop agents, coding swarms and an MCP client and server

Why it matters. New to this directory—not necessarily newly launched. Compare its workflow and requirements with tools in the same category.

Profile context & before you switch

Where it stands out

Supports local models. Supports multi-agent workflows. Lets you bring your own model.

Before you switch

Open-source software on your own hardware. Hardware, electricity, optional connected services and individual model licenses remain separate.

Runs local · linux, macos.

Linux is the main platform. Apple Silicon support is partial; do not assume the Linux desktop, GPU and media workflows are identical on macOS. Windows/WSL2 is not established here.

An alternative Docker core-stack installation is documented. This is not evidence that every agent command runs in a Docker security sandbox.

Compare with BitFun and OpenChamber

Same-category alternatives, matched by execution, platform and capabilities. Open the full comparison for source dates and qualifications.

ProfileGuaardvarkBitFunOpenChamber
Runslocallocallocal
PricingOpen-source software on your own hardware. Hardware, electricity, optional connected services and individual model licenses remain separate.The core code is MIT and the app is free to download; you configure a provider API key and pay that providerFree and open source under MIT and self-hosted; you pay whichever provider the OpenCode agent underneath is configured against
Local modelsSupportedNot listedNot listed
Open sourceMITMITMIT
New to the board· Terminal

Meta's terminal and CI coding agent with approvals, OS sandboxing, MCP and parallel agent workflows

Why it matters. New to this directory—not necessarily newly launched. Compare its workflow and requirements with tools in the same category.

Profile context & before you switch

Where it stands out

Supports headless or CI workflows. Supports multi-agent workflows. Subagents are documented. The workflow engine and session messaging have build, rollout and platform restrictions; session messaging is unavailable on Windows.

Before you switch

Sign in to an eligible Meta plan or use your own Meta API key. Model usage is billed under the applicable plan; check current limits and data-use mode.

Runs local / sandbox · macos, linux, windows.

The reviewed documentation provides a native installer but does not establish a source-code license.

A lasting free inference allowance was not established in the reviewed docs. Temporary credits and promotions must be checked with the publisher.

Compare with Zero and Poolside

Same-category alternatives, matched by execution, platform and capabilities. Open the full comparison for source dates and qualifications.

ProfileMuse CodeZeroPoolside
Runslocal / sandboxlocal / sandboxlocal / sandbox / cloud
PricingSign in to an eligible Meta plan or use your own Meta API key. Model usage is billed under the applicable plan; check current limits and data-use mode.Free and open source under MIT; bring your own provider key or run a local modelFree access to the Poolside Platform with preview Laguna models; paid access through OpenRouter and enterprise self-managed deployments via sales
Local modelsNot listedSupportedSupported
Open sourceNoMITNo
Release log · other recent releases

Version tracking, separate from featured stories.

Loading agent listings… Models and guides are ready.
↑↓ navigate · Enter open · Esc closeModels & guides