Maestro Mojo Maestro MojoYour agents forget everything between sessions. Their work is invisible to the team, and two sessions will happily build the same thing twice. Maestro is the shared board, knowledge, and agent pipeline that fixes that — and you drive it from the AI tool you already use. Bring the AI you already trust. Maestro gives it shared project memory, an engineering workflow, and a cloud workspace that survives the conversation.
Free when you bring your own AI — credits only apply when you use Maestro's AI
Keep the improvisation for tiny changes. When the work needs to survive the chat, Maestro turns it into a shared task, a challenged plan, and a verified build.
Durable context, challenged plans, visible work, and optional execution around the AI you already use.
Index any repo — 18 languages, tree-sitter powered. Trace callers, callees, and transitive dependencies across projects, so the next session starts with impact instead of reconstructing the codebase.
Semantic search across every project's code, docs, patterns, and decisions keeps hard-won context reusable. Ask "does any project already handle SAML?" and get an answer across your entire stack.
Task, scope, and build skills create a shared workspec, a challenged repo-grounded plan, and a verified implementation that survives the chat.
Approve a plan, then Maestro's agents write code, run tests, and open a pull request. Execution is optional; you keep the review gate.
Agents and teammates work the same visual boards. Stages, comments, commits, and client attribution stay visible in Workspaces.
Claim a subdomain or bring your own domain with automatic TLS. Deploy static sites and docs from chat, with cookieless analytics built in.
Bring your own AI. Maestro stores the tasks, plans, knowledge, indexed code, and results you intentionally share — not a copy of your complete chat transcript. Boards, reads, knowledge, code queries, sites, and docs use no credits; AI-backed intake, builds, and initial indexing do. Privacy · Pricing
The shared surface where it all lands. Agents file, build, and comment; your team sees every move in realtime — no status meetings.
Visual tracking boards with sub-items, milestones, priorities, and activity trails. An agent claiming an item and a teammate dragging one are the same move on the same board.
Changes from agents or teammates patch into the open board live, with attribution — "(via Cursor)", "(via Codex)", naming the tool that acted on your behalf.
Publish a board read-only or open to filing, pulse metrics on throughput, workspace docs and skills your agents actually apply.
Ship the site from the same chat — free subdomain or your own domain, automatic TLS, versioned deploys, one-tool rollback.
"Put the docs book on docs.acme.com" is a deploy. Handbooks arrive already designed — docit-powered docs, wikis, and blogs with instant search, chapter nav, RSS, cookieless analytics, and a markdown twin of every page for AI readers.
Two ways in. Start in Claude Code and your terminal is the signup form — or sign in on the web first and use MCP connections, which hands you the right prompt or config for whatever you're using — Claude Code, ChatGPT, Cursor, Zed and more — with your key already in it.
Works with Claude Code · Claude Desktop · Cursor · Codex · Windsurf · Cline · Zed · any MCP client
Run this in your terminal, then restart Claude Code — it reads MCP config only when a session starts. User scope keeps Maestro available in every repo, not just this one.
claude mcp add --scope user --transport http maestro https://maestromojo.com/mcp
First time connecting? In the restarted session, ask Claude to sign you up. (Already have a Maestro key from MCP connections on the web? Skip this — verifying rotates your key and would break that connection.)
Register me with Maestro using you@example.com
Maestro emails a 6-digit code. Hand it straight back:
Verify my Maestro code: 123456
Verifying prints your key and the exact claude mcp add
line with it baked in. Drop the keyless server, run the line Maestro
printed, restart once more — the tools are yours.
claude mcp remove maestro
Open Workspaces, sign in, then open MCP connections for the right prompt or config. Claude Code gets one prompt and a restart — not two prompts. Once connected, say “Confirm my Maestro connection, then sync my Maestro skills.” Approve any skill files it offers; this chat can use the workflow immediately, and saved commands appear next session. Then say “Use maestro-task to turn this into a well-thought-through task: …” The dialog can re-show or rotate your key whenever needed.
Fair question — monday.com, Linear, and roll-your-own agent stacks get asked too. Each is excellent at what it was designed around. None was designed around this: agents and humans doing the same work, on the same board, under the same rules.
GitHub Issues tracks the work humans do in one repo. monday.com runs the work humans do anywhere. Agent frameworks run agents with no shared surface. Maestro is the layer where agents and humans work the same board — same knowledge, same rules, every project. See the full comparison →
Design centers, not scorecards — each of these is excellent at the job it was built for. The question is which job your agents hit first. Written as of August 2026.
Agents without a conductor drift — duplicated work, stale knowledge, nobody sure what changed overnight. Give yours the podium.
Free when you bring your own AI — credits only apply when you use Maestro's AI