Replicas
Run Claude Code, Codex, Cursor, and Opencode coding agents in isolated cloud Linux VMs, triggered from Slack, Linear, or GitHub.
Replicas solves the boring half of the cloud-agent problem: provisioning, harness choice, and attribution. The pitch that lands is Parrot's — same Claude Code, Codex, or Cursor you already use, just not on your laptop. Budget discipline matters more than seat count here, because overage is billed per minute after included usage. Track merged PRs per dollar, not agent minutes.
Verified 2d ago · liveness 71/100 · cite: rightaichoice.com/tools/replicas
- Engineering teams automating recurring bug fixes, refactors, and dependency maintenance
- Teams already delegating from Slack, Linear, GitHub, or GitLab who want agent runs where they work
- Organizations running multiple harnesses (Claude Code, Codex, Cursor, Opencode) in one platform
- Teams needing per-minute attribution by person, model, harness, and credential
- Non-developers wanting no-code automation with a visual builder
- Teams needing on-premise or air-gapped deployment
- Developers seeking real-time interactive pair-programming inside their editor
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Replicas if you want a visual no-code automation builder, an in-editor pairing copilot, or an on-premise/air-gapped install — and if nobody on your team reviews pull requests, it will just generate more code nobody reads.
API and automations are billed by usage on every tier, including the $200/mo Team Full seat and the $50/mo Developer seat, so scheduled and webhook runs add spend on top of the seat price.
Replicas sits above in-editor AI coding subscriptions and below per-seat enterprise agent platforms: $50/mo (billed upfront) for one Developer with 150 hours, $200/mo (billed upfront) per Team Full seat with unlimited manual workspace minutes, plus pay-as-you-go Flex at $0.008/min (Developer) and $0.016/min (Team) for spiky usage. Compare that to buying managed cloud-agent seats per developer elsewhere — the deciding factor is how much of your work runs on automations, since those are billed by
In short
Replicas — Run Claude Code, Codex, Cursor, and Opencode coding agents in isolated cloud Linux VMs, triggered from Slack, Linear, or GitHub. Best for Engineering teams automating recurring bug fixes, refactors, and dependency maintenance, Teams already delegating from Slack, Linear, GitHub, or GitLab who want agent runs where they work, Organizations running multiple harnesses (Claude Code, Codex, Cursor, Opencode) in one platform. Free to start; paid plans from $50/mo.
What's new in Replicas
Checked todayAcross the latest 10 updates: 6 feature updates, 1 launch and 3 news mentions.
Replicas raises $4M seed round
Replicas announced a $4M seed round, per a company blog post by Connor Loi.
How Vela made engineering reviews 3× faster with cloud agents
Customer story: Vela reports 3× faster engineering reviews after adopting Replicas cloud agents.
Introducing Replicas for iOS
Replicas released an iOS app, extending the cloud coding agent beyond the web.
How to use Mercury with Replicas
Integration guide published for using Mercury with Replicas.
How Utility Profit ships code across all departments with cloud agents
Customer story: Utility Profit uses Replicas cloud agents to ship code across departments, not just engineering.
How to use ClickHouse with Replicas
Integration guide published for using ClickHouse with Replicas.
Giving Agents Memory
Replicas published on adding memory to its cloud coding agents.
Role-based access control
Replicas added role-based access control for team permissions in its product.
Code intelligence in every workspace
Replicas rolled out code intelligence features across all workspaces.
Introducing fallbacks + forks
Replicas shipped fallbacks and forks for its cloud coding agents.
What people actually say about Replicas — is it worth it?
We scanned public community sources for Replicas on Aug 26, 2026 and could not establish that the discussion we found is about this tool rather than something else sharing its name. Our own analysis of that scan says the posts were off-subject. Rather than publish a sentiment score built on the wrong subject, we publish nothing here and re-run the scan.
Viability Score
How well maintained and how widely used is Replicas? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Run coding agents in isolated cloud Linux VMs
- Harness-agnostic: Claude Code, Codex, Cursor, and Opencode each in its own VM
- Full computer use — agents drive a real desktop and browser
- Low-latency computer-use stream for watching cloud agents work
- Returns pull requests, replies, screenshots, and screen recordings
- Trigger agents from GitHub and GitLab comments or pull requests
- Trigger from Slack channels by pinging Replicas
- Turn Linear issues into active agent runs
- Dashboard chat with diff watching and plan steering
- Automations and API for scheduled, webhook, or single-call runs
- Warm pools and warm hooks preinstall dependencies and images
- Configurable environments with MCPs, skills, files, and variables
- Persistent memory for cloud agents
- Role-based access control (RBAC) for granular permissions
- iOS app to start and follow agent runs from a phone
About Replicas
Replicas is a cloud coding agent platform. Instead of babysitting an agent in an editor tab, you hand it a task and it runs inside its own isolated Linux VM in the cloud, with your repos, dependencies, and tooling already in place. You kick off work from Linear, Slack, GitHub, GitLab, the web dashboard, a webhook, or one API call — and what comes back is something reviewable: a pull request, a reply, or a screen recording of the agent doing the work. The harness choice is yours. Claude Code, Codex, Cursor, and Opencode each run in their own VM on the same environment primitives, so teams can mix agents task by task or run several in one workspace without forcing engineers onto a new tool. Environments get configured once: warm hooks preinstall dependencies and images alongside MCPs, skills, files, and variables. Agents get full computer use — they drive a real desktop and browser, verify their own work, and hand back screenshots and screen recordings as evidence. Automations and the API fire runs on a schedule or webhook with nobody in the room. Analytics attribute every minute to a source, a person, a harness, a model, the paying credential, and the MCP servers touched. Recent releases added RBAC, persistent agent memory, and an iOS app for starting and following runs from a phone. Named customers include Knowunity (cloud agents produce 50% of its pull requests), Parrot (23.5k automated cloud-agent jobs in July), Moda, Mintlify, Vela, and Utility Profit. It fits teams with mature CI/CD and a real pull-request review process. It is not a no-code automation builder, and it is not an in-editor copilot.
Behind the Verdict
Where Replicas earns its place is teams that already delegate in writing. If bug reports land in Linear and reviews happen on GitHub, there is no new workflow to learn — you tag Replicas where the work already shows up and a pull request comes back. The harness-agnostic design is the part I'd pay for: when one model has a good week and another has a bad one, you switch per task instead of migrating your whole team. Pick it when recurring maintenance — dependency bumps, refactors, test fixes — is eating senior-engineer hours. Knowunity crediting cloud agents with half its pull requests and Parrot logging 23.5k automated jobs in a month are numbers from teams that set up the plumbing first. Do that same setup: warm hooks, environment configs, RBAC, and a review queue that can absorb agent PRs faster than humans write them. Pass when nobody owns the review. Autonomous agents generate pull requests faster than an unprepared team can triage, and a backlog of unreviewed AI PRs is worse than no automation. Also pass if you need air-gapped or on-prem deployment, a visual no-code builder, or real-time pair programming inside your editor. Replicas is asynchronous by design. The closest comparison is running Claude Code or Codex locally in a loop, or a CI-integrated agent. Local keeps context but ties up a laptop and shares your machine's state; CI is deterministic but awkward for open-ended coding. Replicas sits between — sandboxed and asynchronous, with the agent able to install, run services, and drive a browser to check its own work. Watch the meter. Every tier bills usage on top of the seat, and the per-minute rate depends on workspace size — small, large, ultra — with a higher rate after included usage is spent. Automated and API runs consume those minutes too, so a
Researching Replicas? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Replicas actually fits — and what changes day-one when you adopt it.
You wire warm hooks to preinstall your dependencies, images, MCPs and skills, create a shared environment, then tag Replicas in a GitHub pull request comment on a flaky-test bug. The agent boots its own VM, reproduces the failure, opens a PR with the fix, and hands back a recording of the test run.
Outcome: A reviewable pull request with evidence attached lands in GitHub without any developer leaving the review thread, and the run's minutes are attributed to the pull request source and the credential that paid for it.
A Linear issue sits untouched for a week. You convert it into an agent run straight from Linear; Replicas reads the ticket, gathers context and starts implementation in a configured Linux VM while the team works on other things.
Outcome: The issue comes back as a pull request or a reply with questions, which your reviewer either merges or sends back for another pass — one handoff removed from the queue.
You take the $50/mo Developer seat (billed upfront, 150 hours a month) and run refactors and dependency maintenance from the dashboard, watching the diff and steering the plan when the agent drifts.
Outcome: Recurring maintenance gets done off-hours at a fixed seat price, with API and automation usage metered separately so you can see when a scheduled job is the thing driving your bill.
Use Cases
- Delegate bug fixes to a background agent working in an isolated VM
- Automate code refactoring across repositories
- Implement new features with autonomous agents that verify their own work
- Generate screenshots and screen recordings as evidence an agent verified its work
- Start event-driven tasks by tagging Replicas in Slack, Linear, GitHub, or GitLab
- Clear engineering backlog while developers focus on deep work
- Attribute agent minutes by person, harness, model, and credential for cost review
- Run agents on a schedule or webhook with nobody in the room
Models Under the Hood
as of 2026-10-15
Limitations
- Usage-based pricing: Developer is $50/mo including $50 of usage, Team $200+/mo including $200+ of usage, with overage rates starting at $0.004/min for small (2 vCPU, 8 GB), $0.008/min for large (4 vCPU, 16 GB), and $0.016/min for ultra (8 vCPU, 32 GB) workspaces.
- Unlimited manual and automated workspace minutes, custom warm hooks and pools, custom API rates and rate limits, and SOC 2/DPA/SCIM are listed only under Enterprise.
- Higher API rate limits and a static egress IP are listed starting at the Team tier.
as of 2026-10-09
Verification history
We have re-verified Replicas 11 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 11 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Replicas tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Developer
$50/mo
Ideal for
One individual running background agent work regularly and wanting a predictable ceiling rather than metered minutes.
What this tier adds
$50/mo billed upfront for one person, adding 150 hours per month plus API access and the mobile app on top of Flex capabilities.
Team
$200+/mo
Ideal for
An engineering team standardising on cloud agents for recurring refactors, maintenance and backlog work.
What this tier adds
$200/mo billed upfront per Full seat replaces metered manual minutes with unlimited ones, and adds a static egress IP plus auto-upgraded sandbox resources (4 vCPU, 32 GB disk, 16 GB memory) and a shared Slack support channel.
Enterprise
Custom
Ideal for
Organizations whose procurement, security or compliance review gates adoption of agent tooling.
What this tier adds
Custom-priced top tier with unlimited manual and automated workspace minutes, custom warm hooks and pools, custom API rates and rate limits, SOC 2, DPA, SCIM user provisioning and custom security and privacy settings.
Where the pricing makes sense
The company stage and team size where Replicas's pricing actually pencils out — and where peers do it cheaper.
Replicas sits above in-editor AI coding subscriptions and below per-seat enterprise agent platforms: $50/mo (billed upfront) for one Developer with 150 hours, $200/mo (billed upfront) per Team Full seat with unlimited manual workspace minutes, plus pay-as-you-go Flex at $0.008/min (Developer) and $0.016/min (Team) for spiky usage. Compare that to buying managed cloud-agent seats per developer elsewhere — the deciding factor is how much of your work runs on automations, since those are billed by
Setup time & first value
How long it actually takes to get something useful out of Replicas — broken out by persona, not the marketing-page minute.
A single developer on the $50/mo Developer seat can start a first run the same afternoon — connect a repository, trigger from the dashboard or a Slack ping, and review the PR. Teams wiring warm hooks, preinstalled dependencies and images, MCPs, skills and shared environments should budget days rather than hours, and onboarding a second harness or adding Enterprise items (SCIM, DPA, SOC 2) takes
Switching to or from Replicas
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From in-editor AI copilots (Cursor, Copilot): keep your existing harness — Replicas runs Cursor and Claude Code in a cloud VM, so you move the execution location, not the tool.
- →From single-harness cloud agent tools: point Replicas at the same repositories and run Claude Code, Codex, Cursor and Opencode side by side on the same environment primitives.
- →From manual backlog triage in Linear: connect Linear so issues become agent runs instead of tickets waiting for a human.
- →From chat-based agent scripts: replace ad-hoc bots with automations and the API, which fire runs from a schedule, a webhook, or one call.
- →From developer laptops: reproduce your local environment once through warm hooks and images, then let each task boot its own isolated Linux VM.
- ↗To an in-editor copilot: if your team wants real-time interactive pair programming inside the editor, local tools are the fit and the agent output style changes completely.
- ↗To a no-code automation builder: if the work isn't code and reviewers aren't reading pull requests, a visual workflow tool is the better shape.
- ↗To a single-harness cloud agent vendor: if you only ever want one harness, a narrower platform may cost less per seat — you'd lose the ability to mix Claude Code, Codex, Cursor and Opencode.
- ↗To self-hosted agents: export your environment definition as hook and image steps and run agents on your own infrastructure if air-gapped or on-premise deployment becomes a requirement.
- ↗To ticketing-only workflows: if you never want agent-authored pull requests, keep the issue tracker and skip the cloud agent layer entirely.
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “Replicas”, and we withheld 6: 6 could not be judged, because “Replicas” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Replicas.
Official links
Tools that pair well with Replicas
Common stack mates teams adopt alongside Replicas, with the specific reason each pairing earns its keep.
Ellipsis
Run Claude Code and Codex coding agents in managed cloud sandboxes defined by YAML in your repo.
Capy
Capy runs a fleet of cloud coding agents in parallel across Slack, GitHub, and Linear.
OpenHands
OpenHands runs autonomous coding agents triggered by GitHub, Slack, Jira, or schedules to review PRs and fix CI.
Featured Head-to-Head Comparisons
Replicas vs Locus Robotics
Locus Robotics and Replicas serve entirely different domains: warehouse automation vs. AI coding agents. Locus is ideal for high-volume fulfillment centers needing flexible AMR fleets with a RaaS model, while Replicas targets engineering teams wanting to offload coding tasks to background agents. Choose based on your operational focus—physical logistics or software development.
Replicas vs Presto Voice
These tools solve entirely different problems. Presto Voice is a specialized drive-thru voice AI for QSR chains looking to boost revenue via automation and upselling. Replicas is a developer tool for engineering teams that want to offload coding tasks to autonomous agents. Choose based on your domain: restaurant operations or software development.
Replicas vs Truleo
Truleo and Replicas serve completely different domains—law enforcement intelligence versus software development automation. There is no direct competition, so the choice is purely based on your role: if you're in law enforcement drowning in siloed data, Truleo is purpose-built; if you're an engineer wanting to offload coding tasks to autonomous agents, Replicas is the answer. No cross-over possible.
Alternatives to Replicas
View all
Frequently Asked Questions
Categories
Best-of guides
Used Replicas? Help shape our editorial sentiment research.