Last updated: 11/20/2026Last verified: 2026-11-20
Mastra Factory is an open-source, agent-powered software delivery environment where issues from GitHub, Linear or Slack move through intake, triage, planning, building and review, with people answering questions and approving plans and pull requests along the way.
What is Mastra Factory?
Mastra Factory is an open-source environment, described by Mastra as one for combining persistent coding agents, repository workspaces, issue intake, planning, implementation and pull request review in a configurable SDLC. It moves work through the stages intake, triage, planning, building, review and done, taking issues from Linear, GitHub and Slack. Agents run in cloud sandboxes with the repository already cloned, every change comes with a live preview link, and the product is Slack-native: a team can mention Factory in a thread to file work, answer its questions and approve plans. It is not unattended: people answer questions during triage, edit and approve the plan before implementation, and decide on the pull request. The official site states the project is open source under Apache-2.0 (its structured data lists the Apache-2.0 licence and its blog announces "Mastra is now Apache 2.0 licensed!"). Details here were verified from mastra.ai/factory and mastra.ai/pricing on 2026-11-20; the product was not installed or tested.
Key Features
- Six explicit stages: intake, triage, planning, building, review and done, with configurable rules and gates around each one.
- Issue intake from existing tools: the official page lists Linear, GitHub and Slack as sources, and a team connects a GitHub organization and repository, optionally a Linear project, and a model provider.
- Persistent coding agents in cloud sandboxes: every agent gets its own cloud sandbox with repos cloned and ready, and each work item keeps its own session.
- Live preview links: every change ships with a preview link so reviewers click through the real app before opening the diff.
- Slack-native collaboration: mention Factory to file work, answer its questions and approve plans from the thread.
- Human gates: people answer triage questions, review and edit the plan before implementation, and decide whether to merge the pull request.
- Cloud dashboard: teams work from a shared dashboard instead of one person's terminal.
- Optional /goal mode: the page describes the option to implement full /goal mode via subagents, durable background agents and shared state.
- Open source and portable: Apache-2.0 per the official site, deployable automatically on the Mastra platform or hosted on your own infrastructure.
- Local development first: a Factory project uses the Mastra platform for auth, database and sandboxes from the start, while the Mastra server, Factory UI and Studio can run locally until you deploy with mastra deploy.
Best For
Pricing
Official pricing (mastra.ai/pricing, read 2026-11-20) states "Prices per workspace, each month." Free is $0/month, Starter $25/month, Team $250/month, and Enterprise is custom (get in touch). Free includes unlimited agents, users and deployments, 100K observability events, 1 GB egress, 5 connections and 15-day trace retention. Starter adds 1 small dedicated project, 250K events, 10 GB egress, 10 connections and 60-day traces. Team adds 3 projects including 2 medium dedicated projects, a SOC 2 report, 10M events, 100 GB egress, 50 connections and 180-day traces. The same page lists metered overage per tier after that tier's included allowance: memory $0.25/GB-hour, vCPU $0.25/vCPU-hour, connections $0.50/connection, data egress $0.10/GB and observability events $10 per 100K events. Mastra Factory itself is open source, so hosting costs depend on whether you deploy on the Mastra platform or your own infrastructure.
| Plan | Price | What is included |
|---|---|---|
| Free | $0 / month | Unlimited agents, users and deployments; 100K observability events; 1 GB egress; 5 connections15-day trace retention |
| Starter | $25 / month | Everything in Free plus 1 small dedicated project; 250K events; 10 GB egress; 10 connections60-day trace retention |
| Team | $250 / month | Everything in Starter plus 3 projects including 2 medium dedicated projects; SOC 2 report; 10M events; 100 GB egress; 50 connections180-day trace retention |
| Enterprise | Custom | Everything in Team plus custom connections and dedicated supportContact Mastra |
Pros & Cons
Pros
- Connects issue intake, planning, implementation and pull request review in one staged workflow.
- Keeps human input and approval embedded at triage, planning and review instead of presenting the pipeline as unattended.
- Gives each agent a cloud sandbox and each work item its own session.
- Attaches a live preview link to every change, so review can start from the running app.
- Lets teams file work, answer questions and approve plans from Slack.
- Is open source under Apache-2.0 per the official site, with a choice between Mastra platform deployment and self-hosting.
Cons
- Work is gated on people: triage questions must be answered, plans edited and approved, and the pull request decided on, so throughput depends on team availability.
- The official description says a Factory project uses the Mastra platform for auth, database and sandboxes from the start, so deployment is not purely self-contained.
- Platform costs accrue beyond the included allowances: memory, vCPU, connections, egress and observability events are all metered per tier.
- Published limits cover projects, connections, events, egress and trace retention; model-provider and sandbox resource limits are not stated on the pricing page.
- Live previews are described as part of the delivery flow, and preview compatibility across application types is not documented on the pages read for this overview.
FAQ
Details
Platform
Features
- Staged lifecycle from issue to reviewed pull request: intake, triage, planning, building, review and done, with configurable rules per stage
- Persistent coding agents work in cloud sandboxes with the repository already cloned, and each work item keeps its own session
- Every change comes with a live preview link so you click through the app before opening the diff
- People answer the agent during triage, edit the plan before implementation and decide on the merge in review
- Slack-native: mention Factory in a thread to file work, answer its questions and approve plans
- Open source under Apache-2.0, deployable automatically on the Mastra platform or hostable on your own infrastructure
Languages
Known limitations
- A Factory project uses the Mastra platform for auth, database and sandboxes from the start, even though the Mastra server, Factory UI and Studio can run locally
- Platform plans are billed per workspace per month: Free $0, Starter $25, Team $250, Enterprise custom, with memory, vCPU, egress and observability events metered beyond each tier
- The Free tier includes 100K observability events, 1 GB egress, 5 connections and 15-day trace retention, and a dedicated project only starts on Starter
- It is not unattended: someone still answers questions in triage, approves or edits each plan and decides on the pull request
- It needs a connected GitHub organisation and repository, optionally a Linear project, and a model provider before it can pick up work








