v0.15.1 — Watch your agent think

0.15.0 was a big release. It rebuilt the engine that runs your agents, and it left one thing
feeling worse than before: you'd send a message and then stare at a spinner until the whole
answer landed at once. Is it working? Is it stuck? You couldn't tell.

This update fixes that, adds a gauge you'll wish you'd always had, and cleans up a couple of
rough edges.

Replies show up as they're written

You now see the agent's reply come in while it's writing, the same way you would in a chat app.
The order of the conversation makes more sense too. If the agent says "let me check the docs"
before looking something up, that line gets its own card, then you see the lookup, then the
answer. Before, all of that was squashed into one card at the end.

A meter that tells you when the agent is running out of room

Every AI model can only keep so much of a conversation in mind at once. Go past that limit and
it quietly starts forgetting the beginning: the file you mentioned an hour ago, the decision you
made together earlier. Until now you had no way of seeing it coming.

Under the message box there's now a small bar, like "Context ▰▰▰▱▱▱ 41% of 32k tokens." It
turns yellow as it fills and red when it's nearly full. When it's getting high, type
/compact. The agent sums up the conversation so far and carries on from that summary, with
plenty of room to spare.

The colors stay readable in every theme, a full bar also shows ⚠ so you don't have to rely on
color, and screen readers read it out. Cloud models handle their own memory, so for those the
meter just shows how many tokens you've used.

Small fixes

Reply speed is back. The line under each reply shows how fast the model answered again.
It went missing in 0.15.0.
The "Add tag" button isn't cut off anymore. In the Manage tags dialog, the button was
sliding off the edge of the window. It fits now.

Under the hood

This release includes MandoCode CLI v0.15.1.
Everything is in the full changelog.

Install

Download MandoCode.Desktop-v0.15.1-win-x64.zip from this release, extract all of it
into a folder, and run MandoCode.Desktop.exe. Everything you need is in the zip, including
.NET. The only other thing Windows needs is Microsoft Edge WebView2, which most PCs already have.
You'll still need a model provider such as Ollama, and Ollama's cloud models need a
subscription.

Already on 0.15.0? MandoCode Desktop will let you know the update is ready within a day.

v0.15.0 — A new AI foundation and a more connected workspace

MandoCode Desktop now runs on Microsoft Agent Framework (MAF), replacing Semantic Kernel
in the engine. The migration brings conversations, tools, and approved plans onto a shared
agent and workflow foundation. Desktop's notes assistant, snapshot summaries, and skill author
also move off Semantic Kernel. Your existing models, providers, skills, MCP servers, and
approval controls continue to work.

Plans you can steer and resume

Review and edit steps before execution, resume unfinished plans after restarting, and choose
whether to retry, skip, cancel, or approve a revised approach when work fails. Completed steps
stay settled. Use /plan <goal> when you want a plan explicitly; automatic planning now looks
at the shape of the task rather than just the length of your message. Steps carry acceptance
criteria and plans include final testing, replacing the separate model verifier for every step.

A built-in browser—for you and your agents

Browse websites directly inside MandoCode Desktop. Open tabs, enter URLs, and navigate with
back, forward, and reload controls. Ask the agent to explain the page you are viewing or help
fill out a form, including forms embedded in another page. Your request stays tied to that tab,
even if you switch tabs while the agent works.

Watch commands run and preview the results

Live command output: each agent gets a read-only output tab showing commands, working
folders, output, and completion status. A terminal indicator points you to unseen activity.
This displays captured command output; it does not introduce a persistent agent shell.
File and app previews: inspect files beside the conversation, edit text with save
protection, read PDFs, and preview local development sites. Vision-capable models can receive
screenshots; text-only models are told when they cannot inspect an image.

Let your agents work together

Mention another open agent with @ to ask questions or hand off work. You can keep talking
with your agent while the other works, ask for progress, and receive the result when it finishes.
Named split panes and drag-and-drop help you arrange their conversations.

A workspace that remembers your preferences

Save settings independently for each agent while keeping global defaults for new ones. Pin
frequently used models, organize Skills and MCP servers with tags and filtered bulk actions,
and compact a long conversation with /compact. Completed work folds into expandable activity
summaries. Nine new themes, themed backgrounds, and the Midnight Ramen default background join
improvements to text and status-card readability.

Fixes you will notice

Deleting notes, snapshots, or history keeps the rest of the list in place.
Restored conversations retain their model choices without repeated context-window notices.
Changing an agent's project folder keeps its conversation.
Exiting split view brings the agent tabs back.
Routine cloud model switches no longer repeat subscription notices; failures still explain
when a subscription is required.
Token totals use provider-reported usage, and plans with skipped or failed steps report
partial completion instead of an unqualified success.

See the full changelog
for the complete list. This release pins the engine to
MandoCode CLI v0.15.0,
commit e67578251c5716a6ede192a2e3e82e7dfda7c8f0.

Install

Download MandoCode.Desktop-v0.15.0-win-x64.zip from this release, extract all its contents
into a folder, and run MandoCode.Desktop.exe. The Windows x64 package includes .NET;
Microsoft Edge WebView2 Runtime is the only separate system runtime dependency. Model access
still requires a configured provider, such as Ollama; Ollama cloud models require an active
cloud subscription.

First-five-minutes polish, straight from watching 0.14.0's fresh-machine debut.

Your agents have names now. New tabs open as Morphy, Kernel, Cloud — drawn from a 500-name deck — instead of Agent 1, Agent 2. Prefer numbers? One toggle in Settings → Behavior, and anyone who'd already chosen numbering keeps it.
The "can't reach Ollama" screen leads with the fix. On a machine without Ollama, the first option is now Install Ollama for me, not Change the Endpoint URL. The message explains what's happening in plain language ("Ollama isn't running on this computer — it may not be installed yet") instead of leading with a raw socket error.
Honest cloud pricing guidance. ollama.com cloud models require an account with an active cloud subscription — without one, requests fail with 403 Forbidden, which used to surface as a raw error that looked like the app breaking. The requirement is now stated whenever you pick a cloud model, and a 403 gets a clear explanation card with the two real fixes: subscribe, or switch to a free local model.

Install

Download MandoCode.Desktop-v0.14.1-win-x64.zip below, extract the whole zip, run MandoCode.Desktop.exe. Self-contained — no .NET required; WebView2 is the one system dependency (preinstalled on Windows 11). Updating from 0.14.0: extract over your old folder or anywhere new — your settings and sessions live in ~/.mandocode and carry over.

Full details: CHANGELOG.md

The first public release of MandoCode Desktop — a Windows desktop home for local AI coding. Run models on your own machine through Ollama (or free cloud models with an Ollama sign-in), with no API keys and nothing leaving your computer that you didn't send.

Highlights

Multi-agent tabs. Every tab is an independent agent: its own model, its own conversation, its own settings. Run a local model and a cloud model side by side on the same project.
Nothing gets lost. Conversations persist across restarts, closed sessions are searchable in an archive, and context snapshots let you carry a conversation's memory into a different model — or hand it to a fresh agent.
Small local models are genuinely usable (engine 0.14.3). The context window rides on every request — it works with the Ollama desktop app, no restarts — conversations compact themselves before they overflow, and the window floor is sized so small models keep real working memory.
Notes with an AI assistant, MCP server support (Claude-Desktop-compatible config), file @-tagging, diff approvals for every write, themes with custom chat backgrounds — and yes, a music player.

Install

Download MandoCode.Desktop-v0.14.0-win-x64.zip below
Extract the whole zip anywhere
Run MandoCode.Desktop.exe — first-run setup walks you through installing Ollama and picking a model

Self-contained: no .NET install required. The one system dependency is WebView2 (preinstalled on Windows 11 and most Windows 10 machines). The app checks for new releases and will let you know when an update is available.

Full details: CHANGELOG.md · Engine: MandoCode CLI 0.14.3

Releases: DevMando/MandoCode.Desktop

Release list