Skip to content
K

Skills & Agents

Monday Morning ships with a curated library of skills — the /mm: command set — and a roster of agents that do the heavy lifting behind those commands. Both are managed from the desktop app’s Settings: the Skills section lists what’s installed and opens a browsable catalog, and the Agents section lets you create and edit custom agents. Everything lives with your workspace as files an agent can read.

Agents come in two flavors. Workflow agents drive the spec pipeline end-to-end. Specialist agents are discipline-scoped experts you can hand a focused task to without writing a prompt from scratch.

These agents run the spec lifecycle: shaping requirements, writing the spec, breaking it into a task list, implementing, and verifying. They are invoked by the /mm: skills that orchestrate the pipeline, and several run in their own sessions so stages can proceed in parallel.

AgentWhat it does
spec-initializerScaffolds a new spec folder and its starting files
spec-shaperGathers requirements through targeted questions and visual analysis
spec-writerAuthors the spec from the shaped requirements
spec-verifierReviews the spec for completeness and consistency
tasks-list-creatorBreaks the spec into an implementable task list
implementerImplements the tasks against the codebase
implementation-verifierChecks the implementation against the spec and tasks
adversarial-verifierStress-tests the work, hunting for gaps and regressions
product-plannerDefines product vision and roadmap during planning

The flow is roughly: shape → write → verify spec → task list → implement → verify implementation, with the adversarial pass as an extra layer of scrutiny.

Specialist agents are the mm-* agents, each scoped to one discipline, so you can route a review, audit, or fix to the agent best suited for it.

AgentWhat it does
mm-ui-ux-reviewerReviews UI and UX for usability and flow issues
mm-visual-designerHandles visual design and styling work
mm-integrationWires integrations end-to-end across the frontend/backend boundary
mm-a11y-auditorAudits for accessibility issues
mm-debuggerDebugs and traces issues to root cause
mm-qa-engineerPlans and runs QA and testing
mm-perf-optimizerProfiles and optimizes performance
mm-security-auditorAudits for security vulnerabilities

When a session fans work out to agents like these, the fan-out is visible on the stage: the stage session’s sub-agents column lists each sub-agent with its prompt, model, state, and per-sub-agent token usage — so a dispatched specialist is something you can watch, not a black box.

Skill management lives in Settings → Skills — Settings is a stage takeover in the desktop app, not a separate window. From the Skills section you can:

  • Browse the skills installed for the current project
  • Open the catalog — an inline skill catalog subview for browsing and installing skills (no modal, no external registry account)
  • Import a skill from a URL or a local file with /mm:import-skill

Because configuration is per project, two workspaces can carry entirely different skill sets, and changing one project’s skills doesn’t affect another.

Settings → Agents lists your custom agents and lets you create, edit, and delete them through an inline edit subview. Custom agents sit alongside the shipped roster, so a team can codify its own specialists the same way the built-ins are defined.

The skill and agent surfaces are backed by the universal MCP verbs plus one import tool:

ToolPurpose
mm_create (entity: "skill")Create a new skill
mm_import_skillImport a skill from a URL or local file (opt-in imports tool group; always dispatchable by name)
mm_list (entity: "plugin")List installed plugins
mm_get (entity: "plugin")Inspect a specific plugin

See the MCP Tools Reference for full schemas.