Maskin is an open-source workspace for teams and AI agents to turn signals into decisions, shipped work, and measured outcomes. Self-hosted or managed, with Slack, GitHub, Linear, Intercom, PostHog, and HubSpot.
Maskin is an open-source workspace for companies that want humans and AI agents to operate in the same system. It is built around shared loops where signals become decisions, decisions become shipped work, and the results are fed back into the next cycle.
The product is aimed at teams that already use tools like Slack, GitHub, Linear, Intercom, PostHog, and HubSpot, and want those systems to feed a more visible process for deciding what to build, shipping it, and learning from the result. Maskin can be self-hosted or run as managed hosting, and the site also positions it as MCP-native with bring-your-own-model support.
Maskin centers work around a shared workspace where customer signals become decisions, decisions become shipped work, and outcomes feed the next cycle. That keeps context attached to the work instead of scattered across private chats.
The product is built around loops such as insight → bet → task → knowledge, so teams can turn a signal into a scoped action and then capture what was learned. The homepage shows this with an onboarding drop-off example that ends in an updated outcome record.
The marketplace includes ready-made loops for competitive intelligence, lost deals, churn and expansion, customer feedback, and market signals. Each template wires agents, objects, and handoffs for a specific business process.
Maskin supports defined agent roles like Feedback Monitor, Bet Strategist, Developer, Intel Monitor, and CS Advisor. The site frames these roles as persistent jobs rather than one-off chat prompts.
Agents connect to existing tools through integrations and MCP tools, with Slack, GitHub, Linear, Intercom, PostHog, and HubSpot called out on the site. The product is positioned to work with an existing stack rather than replace it.
The platform is open source under Apache 2.0 and supports bring-your-own-model setups with Anthropic, OpenAI, or local Ollama use. The site also says it can be self-hosted or managed and offers EU and US data residency.
Use the competitive intelligence loop to watch competitor changelogs, pricing pages, and job postings. Maskin routes the resulting signal to the relevant team, such as turning feature gaps into product work or battlecard updates for sales.
Feed closed-lost reasons from HubSpot into the lost-deals workflow so patterns can be synthesized into roadmap bets. The site says this process can weight insights by ARR and close the loop when the team ships.
Use the churn and expansion loop to watch for usage drops, login gaps, or error spikes. When a risk pattern fires, the right person gets an alert with context so they can intervene before accounts cancel.
Route customer feedback from Slack or Intercom into a loop that turns a complaint into an insight, then a bet, then shipped work. The site also says the customer who raised the issue can be notified when the fix goes live.
Compose a custom loop when the team already runs a repeatable process that is not covered by the templates. The product lets teams define agent roles, objects, and handoffs so the workflow can run as a shared system.
Maskin is described as an open-source, MCP-native workspace where humans and AI agents close loops together. The site says it is available as self-hosted or managed hosting, so teams can choose how they run it.
The docs say to start with the overview, then explore the concepts, MCP tools, or self-hosted setup. The homepage also suggests entering a problem, notes, or a doc so Maskin can draft a scoped bet and start a workflow.
The source describes a shared workspace for humans and AI agents. Humans review and approve bets, agents handle roles such as monitoring, strategizing, and shipping, and the workspace keeps the decision thread and outcome together.
The homepage and docs indicate it can be self-hosted or run as managed hosting. The site also says it supports EU and US data residency and lets users bring their own model key, including Anthropic, OpenAI, or local via Ollama.
The homepage shows workflow templates for competitive intelligence, lost deals to roadmap, churn and expansion, customer feedback, market signals, and custom loops. It also lists integrations and MCP tools for Slack, GitHub, Linear, Intercom, PostHog, HubSpot, and more.
