AgentMash
AgentMash is an open-source multiplayer coordination layer for teams' AI coding agents. It gives agents shared awareness of teammates' in-progress edits, claims and task context so they can adapt before creating file or dependency conflicts.
At a Glance
- Software development teams
- Startups and small-to-mid-size companies using AI coding agents
- Engineering organizations coordinating multi-agent work on shared repositories
- Enterprise engineering teams requiring self-hosting and private infrastructure
AI Tools by AgentMash
(1)
AgentMash.dev
AI Agent Conflict Detection
Discussions
No discussions yet
Be the first to start a discussion about AgentMash
Latest News
agentmash-cli documentation update covering Codex hook approval
AgentMash updates page announced a guided Lobby tour by Mash
AgentMash launched on stage at AI & Digital Bridge in Astana, according to co-founder Alim Dauletbek
Added approved prompts to a teammate's agent
Products & Services
Hosted coordination room for a repository: agents report edits, prompts, claims and commits; before edits, hooks and adapters check for overlapping files or changed dependencies and inject an advisory.
Desktop app for macOS and Windows that runs developers' existing coding agents side by side, with live team view, conversations, room chat and approved teammate-agent requests.
MIT-licensed open-source npm client installed with npx agentmash login; connects local agents, installs hooks/adapters, exposes status, claims, MCP tools, digests and a git post-commit reporter.
A set of 22 ready-made Claude Code subagents organized across lead, planning, build, check, maintain and creative roles.
Market Position
AgentMash positions itself as an open-source shared-context and coordination layer for AI coding agents, differentiated from manual file-ownership conventions, isolated agent sessions and discovering conflicts only after a merge. It supports many commercial and open-source agents while keeping model subscriptions separate and developer-controlled.
Leadership
Founders
Alim Dauletbek
Co-founder and CEO of AgentMash. Based in Astana; previously CTO of Jobee, a teen gig-work marketplace, lead software engineer at Astana Hub, and co-founder/director of an auditing and automation studio.
Damir Iskakov
Co-founder and CTO of AgentMash. He identifies as CEO/co-founder of Pragma automation and has worked as a full-stack engineer at Astana Hub.
Executive Team
Alim Dauletbek
Co-founder and CEO
Astana-based software engineer; previously CTO of Jobee, lead software engineer at Astana Hub and co-founder/director of an auditing and automation studio.
Damir Iskakov
Co-founder and CTO
Full-stack engineer and CEO/co-founder of Pragma automation; described the product and its conflict-prevention benchmarks on Product Hunt.
Founding Story
The founders encountered a recurring problem when running Claude Code, Codex and other coding agents on the same repository: agents could not see one another, so overlapping edits and broken interfaces were discovered only at merge time. AgentMash was started to make agents aware of teammates' work before an edit, allowing the agents and developers to resolve overlaps earlier rather than relying on manual file partitioning or post-merge repair.
Business Model
Revenue Model
Per-developer subscriptions for hosted team coordination, with agents never counted. The product also sells an annual enterprise plan with self-hosting and a private Docker image; customers bring and pay for their own AI-agent subscriptions.
Pricing Tiers
Up to 2 developers; 1 private repository, unlimited public repositories, 7 days of history and unlimited agents.
Unlimited private and public repositories, up to 90 days of history and unlimited agents; first month free is advertised for monthly billing.
Annual contract, invoiced; self-hosting, private Docker image and enterprise-scoped terms.
Target Markets
- Software development teams
- Startups and small-to-mid-size companies using AI coding agents
- Engineering organizations coordinating multi-agent work on shared repositories
- Enterprise engineering teams requiring self-hosting and private infrastructure
- Multiple developers running AI coding agents concurrently on one repository
- Preventing same-file and semantic/API conflicts before merge
- Coordinating parallel feature work through claims and shared task context
- Keeping coding agents aware across branches, worktrees and different AI tools
- Delegating read-only investigation to a teammate's agent with approval
- Teams that want hosted coordination or enterprise self-hosting without buying model usage
- 1609 Plus
- Soup