SkaleAgents

What it does

Writes code

NoThe tools review supplied code and infrastructure text and return findings, remediations and validation steps; nothing on the site claims the product authors new application code from a description.These workflows produce findings and recommendations for you to review. Submitting an audit or opening a Watchtower session does not itself deploy a…skaleagents.com ↗

Debugs and fixes

PartlyThe SRE Triage bot diagnoses reliability failures from supplied telemetry and names a first safe change, but the fix is left for the engineer to apply through their own process.Turns supplied telemetry, logs, and architecture notes into a reliability review: errors, failed jobs, timeouts, missing signals, and the first safe…skaleagents.com/bots ↗

Reviews code

YesReviewing someone else's source or configuration is the core job: an MCP tool reviews individual snippets, and the DevSecOps bot explicitly accepts a pull request diff as input.review_architecture Review an individual source snippet or infrastructure configuration for security, reliability, and cost concernsskaleagents.com/mcp-config ↗

Whole-codebase work

PartlyAn application-wide review maps related source files that the editor sends, but there is no repository indexing, and scheduled audits reach only one file in a public GitHub repository.Scans read the file text your editor sends. They do not connect to a live cloud account. Pulumi programs are scanned as text, so computed values and…skaleagents.com/mcp-config ↗

Ships it

NoThe product stops at findings by design: submitting an audit does not deploy anything, and the vendor tells buyers to apply fixes through their own deployment process.Validate proposed fixes and apply them through your own development and deployment process.skaleagents.com ↗

Done-for-you service

NoThis is self-serve software with organization plans an owner picks; the vendor supplies prompts and tools rather than people who run reviews for you.SkaleAgents gives developers reusable specialist prompts and tools for reviewing code and infrastructure.skaleagents.com ↗

Not established: Writes tests

How much it does unattended

Runs autonomously

PartlyScheduled audits run unattended on a cron expression with webhook alerts, but every result is a finding a person must read and act on; nothing is applied automatically.Create a scheduled audit with a cron expression and timezone. An optional webhook receives completion or failure alerts.skaleagents.com ↗

Multi-agent

PartlyOne audit job can be scoped to several specialist reviewers at once, which is parallel specialists on a shared input rather than agents negotiating or delegating between themselves.Sign in, paste your IaC, and select DevSecOps, SRE, MLOps, or scaling specialists. Add telemetry JSON when it helps.skaleagents.com ↗

Agent permissions

PartlyPermissions are narrow by construction rather than configurable: the agent sees only text you send, never a cloud account or private repository, and cannot change a live system.Schedules can check one file in a public GitHub repository; private repositories and cloud accounts are not connected.skaleagents.com ↗

Which models it runs on

Claude

PartlyClaude is supported as the assistant you bring: bot pages hand off to Claude, and Claude Code and Claude Desktop connect to the MCP server, but you do not select a model inside SkaleAgents.Bot pages offer handoffs to Grok, SuperGrok, and Claude. Codex opens separately, and Copy prompt lets you paste the instructions into any assistant.skaleagents.com ↗

GPT

PartlyChatGPT and Codex are named as supported MCP clients, so GPT models reach the tools through OpenAI's own products; the vendor never states which model its web audit runs on.For tool calls, connect Cursor, Claude Code, Claude Desktop, ChatGPT, or Codex to our hosted MCP server through OAuth.skaleagents.com ↗

Model choice

PartlyBots are instruction sets you run in whichever assistant you already pay for, so the model is your choice there; the model behind the hosted web audit is never disclosed.A bot is a set of instructions, not a separately trained model.skaleagents.com ↗

Not established: Open models

Where you use it

In your editor

YesCursor is a first-class surface with a documented .cursor/mcp.json snippet, and Claude Desktop connects the same way; the review lands in the editor conversation.Save this in .cursor/mcp.json at your project root. Open Cursor Settings, then Tools & MCP. Enable skaleagents and use Connect to complete OAuth.skaleagents.com/mcp-config ↗

On the command line

PartlyThere is no SkaleAgents CLI of its own, but terminal clients are documented: Claude Code and Codex are configured and re-authenticated with their own shell commands.For Codex, run codex mcp login skaleagents.skaleagents.com/mcp-config ↗

In your pipeline

NoNothing acts on a pull request or runs in a pipeline; the only repository reach is a scheduled check of a single file in a public GitHub repository, with private repos excluded.Schedules can check one file in a public GitHub repository; private repositories and cloud accounts are not connected.skaleagents.com ↗

In a browser

YesA hosted web app carries the account workflows: audit jobs from pasted infrastructure text, plus shared Watchtower sessions where a team inspects the findings together.Submit infrastructure text as an audit job, then inspect the findings in a Watchtower session with your team.skaleagents.com ↗

Whose machine it runs on

Self-hosted

PartlyThe MCP server is hosted by the vendor and the setup page says nothing is installed locally, yet the homepage describes the client launching an npx package locally; the two pages disagree.Your client starts @skaleagents/swarm locally with npx; Node.js 20 or newer is required.skaleagents.com ↗

Not established: Open source

What it costs to run

How it meters

PartlyBilling is a flat monthly price per organization with audit and schedule quotas attached, so the metered unit is an audit run rather than a seat, a token or a credit.$49 / month. Room for regular audits across an active organization. 200 audits / month. 10 automatic schedules.skaleagents.com/pricing ↗

Free tier

YesA standing $0 plan gives two audits a month, and the bot directory, prompt library and check reference are readable with no account and no payment at all.$0 / month. Start reviewing infrastructure on your own schedule. 2 audits / month. 0 automatic schedules.skaleagents.com/pricing ↗

MCP server

YesA hosted MCP server at one URL exposes four named review tools over OAuth, so another agent can call the scans and architecture reviews directly.Connect your assistant to the hosted SkaleAgents MCP server. Sign in with Google and approve access through OAuth. No API keys or local server to…skaleagents.com/mcp-config ↗

Bring your own key

PartlyInference for the bot workflows is yours: you run the instructions in an assistant you already pay for, and the vendor states its plans do not cover that spend. No key field or own-cloud option is documented.Using a prompt in another AI assistant is subject to that provider's pricing and usage limits; SkaleAgents does not include a subscription to those…skaleagents.com ↗

Not established: API access

Buying it for a team

A company can buy it

YesThe entire rate card is organization-level: an owner buys Starter, Pro or Max for an organization they manage, with membership and invitations handled in the dashboard.Starter, Pro, and Max are organization plans. An owner chooses a plan for an organization they manage.skaleagents.com/pricing ↗

Seat model

NoNo tier is priced or limited per seat: each published plan states only a monthly price, an audit allowance and a schedule count, with no user count, floor or ceiling.$19 / month. A steady review rhythm for smaller projects. 25 audits / month. 1 automatic schedules.skaleagents.com/pricing ↗

Pooled budget

YesAllowances belong to the organization rather than to each member, and the Enterprise tier negotiates those organization-level limits, so members draw on one shared audit budget.Custom quote. A plan built around your security and delivery needs. Audit and monitoring limits are agreed with the organization.skaleagents.com/pricing ↗

Admin controls

PartlyOwners control who is in an organization and creators can keep bots private, but no published control constrains what the agent itself may read, call or do.Organization owners control membership and should invite only people authorized to see that organization's content.skaleagents.com/terms ↗

Single sign-on

PartlySign-in is Google OAuth only, and MCP authorizations are tied to that Google-backed account; no SAML or enterprise OIDC federation is offered on any published tier.MCP clients are authorized through OAuth and tied to your Google-backed account.skaleagents.com ↗

Not established: Audit log

What happens to your code

Opt out of training

PartlyThe licence you grant is limited to running and protecting the service, which excludes model training, but there is no explicit training statement or opt-out switch to point a buyer at.You grant us permission to host, process, and display that material only as needed to provide and protect the service.skaleagents.com/terms ↗

Data residency

PartlyProcessing is disclosed but not selectable: hosting runs on Railway and the policy reserves United States or other provider locations, with no region choice offered to a buyer.Information may be processed in the United States or other countries where these providers operate.skaleagents.com/privacy ↗

Getting out

PartlyExport and deletion exist as rights exercised by email plus some in-product purge controls, not as a self-serve export button; backups may retain data after deletion.You can edit some account and organization information in the dashboard, revoke connections and tokens, and request access, correction, export, or…skaleagents.com/privacy ↗

Certifications

NoThe vendor states plainly that enterprise compliance certification is not promised, and no SOC 2, ISO 27001 or audit report is claimed anywhere on the site.Cloudploy deployment handoff, Crontinel telemetry integration, and enterprise compliance certification are not promised by the workflows on this page.skaleagents.com ↗