Anthropic Claude Code vs Lovable (2026)

MCP servers connect the agent to tools, databases, APIs, issue trackers and monitoring systems at local, project and user scopes, with OAuth 2.0 sign in for remote servers and managed allowlists, denylists and per tool ask or block settings for organizations, alongside built in file, search, shell and web tools.

Sourcecode.claude.com/docs/en/mcpread 2026-09-29

Connectors sit in a single catalog, with Lovable handling authentication, credentials and requests. There are three connection types, app connectors, chat connectors and app user connectors, where each person using the published app connects their own account, and each connector's page says which types it supports.

About ninety integrations have their own pages, spanning Salesforce, HubSpot, Pipedrive and Zoho CRM, Snowflake, BigQuery, Databricks, Redshift, ClickHouse and Microsoft Fabric, Shopify, WooCommerce, PrestaShop and Lightspeed, Stripe, Chargebee, QuickBooks and Xero, Slack, Twilio, Telegram and Resend, Workday and Ashby, Aikido and Wiz, Notion, Airtable and Contentful, and AWS S3. An any API route, a create connector path, custom MCP servers and MCP registries cover anything not listed.

Sourcedocs.lovable.dev integrations introduction, app-connectors, chat-connectors, app-user-connectors, any-api, create-connector and integration catalogread 2026-08-31

Subagents run in their own context windows with their own prompt, model, tools and permission mode, can run in parallel or in the background, and include built in Explore, Plan and general purpose agents; agent teams let a lead session spawn and supervise coordinated sessions that message each other, and /batch splits large changes into parallel background subagents after a plan is approved.

Sourcecode.claude.com/docs/en/sub-agents and docs/en/commandsread 2026-09-29

Build mode, formerly Agent mode, works on its own and takes a task end to end. It understands intent, explores the codebase for context, applies changes across files and resolves issues during development, with visible tasks showing progress and results reviewed before moving on.

Plan mode handles decisions, and the user can switch between the two at any time. Subagents are a separate feature, and cross project referencing lets one project draw on another. Skills apply reusable instructions, code mode and the project editor expose generated files for direct editing, and drafts and project history support iteration.

Sourcedocs.lovable.dev agent-mode, plan-mode, subagents, cross-project-referencing, skills, code-mode, projects/editor, drafts and projects/history pagesread 2026-08-31

The Claude Code GitHub Action runs in automation mode on any GitHub event, such as a pull request being opened or a cron schedule, without anyone mentioning it, and Code Review reviews every pull request automatically; beyond that the agent answers @claude mentions and runs from the terminal, desktop, IDEs, the web and headless scripts.

Sourcecode.claude.com/docs/en/github-actionsread 2026-09-29

Lovable can check a project on a schedule, daily or weekly at a chosen time and optionally only after a minimum number of edits, reviewing the app code for broken functionality and recent visitor errors, then alerting by email and listing findings to fix in the project chat. So the builder agent wakes on a schedule without anyone asking. Other routes in are the web chat, a Slack app, mobile and desktop apps, a Telegram route, the Lovable API and the Lovable MCP server. Monitoring is in beta, on Pro, Business and Enterprise.

Sourcedocs.lovable.dev project-monitoring page and llms.txt indexread 2026-09-29

The agent grounds itself in the customer's repository by reading and searching it directly on each run, with project instructions from CLAUDE.md or AGENTS.md and path scoped rules, and MCP connections reach the customer's databases, issue trackers and internal APIs. Context is gathered fresh on each run, and Anthropic does not describe an index, graph or embeddings layer that persists between runs.

Sourcecode.claude.com/docs/en/memory and docs/en/mcpread 2026-09-29

Knowledge holds persistent instructions and context for the agent, defined once and always included, at two levels. Workspace knowledge holds shared rules across all projects, such as coding standards, preferred libraries and naming conventions, and project knowledge holds the app's purpose, database schema, architecture decisions and domain terminology. Workspace knowledge is a text field, unlike Skills, which load on demand.

Project and workspace knowledge plus the customer's code and components are the main context sources, and Lovable does not describe reading instruction files from the repository or from outside sources. Design systems are Lovable projects that export reusable components, colors, fonts and styling rules, which other projects connect to and inherit, with immediate propagation, an order of precedence across several connected systems, and no versioning. Cross project referencing and about ninety integrations supply more material for each request.

Sourcedocs.lovable.dev features/knowledge, design systems FAQ and knowledge context-sources FAQread 2026-08-31

Auto memory, on by default, is written by Claude itself as it works (user, feedback, project and reference notes), scoped per repository on the user's machine, loaded into every session through a MEMORY.md index, kept outside the transcript retention sweep until the user or Claude edits or deletes it, and browsable and editable through /memory. CLAUDE.md files are standing instructions the user writes, separate from this memory.

Sourcecode.claude.com/docs/en/memoryread 2026-09-29

Projects are durable objects with a history page recording changes, drafts keep unfinished work, and project chat persists. A Knowledge feature holds project configuration, and cross project referencing lets one project draw on another. Build mode explores the codebase for context on each task instead of carrying learning between tasks, and skills apply reusable instructions the customer writes, which the agent does not update.

Sourcedocs.lovable.dev projects/history, drafts, projects/chat, knowledge, cross-project-referencing, agent-mode and skills pagesread 2026-08-31

In Manual mode the agent starts read only and asks before it edits files, runs tests or executes commands that modify the system, with approve once or allow from then on, fail closed matching for unmatched commands and a working directory write boundary; auto mode sends actions to a separate classifier and organizations can turn it off, and managed settings enforce permissions above user scope.

Sourcecode.claude.com/docs/en/securityread 2026-09-29

Plan mode investigates the project and writes a plan the user reviews, edits and approves before any code is written, and Lovable says Plan mode never modifies code. It is one of three modes the user picks, and Build mode applies changes directly, so the approval sits in an optional mode, not in the build path. Project monitoring findings are fixed only when the user asks. Workspace admin controls, integration admin controls, groups and project visibility settings limit who can do what.

Sourcedocs.lovable.dev plan-mode and project-monitoring pagesread 2026-09-29

A SOC 2 Type 2 report and an ISO 27001 certificate sit in the Anthropic Trust Center, and Claude Code uses a permission based architecture, sandboxed shell commands with filesystem and network isolation, command injection detection, trust verification for new codebases and MCP servers, keychain credential storage, managed settings for organization enforcement and a HackerOne disclosure program.

Sourcecode.claude.com/docs/en/securityread 2026-09-29

Platform controls include a trust center, SAML single sign on, SCIM provisioning, two factor authentication, workspace identity and identity reuse, groups and people access controls, integration admin controls, project visibility settings and Enterprise audit logs.

Enterprise materials describe centralized identity, granular access controls and security oversight across the workspace. Business tier has a data opt out setting, and secrets and build secrets are kept separate from generated code.

Security scanning of the generated app is separate, through a security center, security insights, sensitive data scanning and Aikido and Wiz integrations.

Sourcedocs.lovable.dev trust-center, saml-sso, business/sso, business/scim, two-factor-authentication, workspace-identity, groups, people, admin-controls, business/data-opt-out and secrets pagesread 2026-08-31

OpenTelemetry exports metrics, events and optional traces, with a tool decision event for every permission accept or reject and a tool result event for every completed tool call, and OTEL_LOG_TOOL_DETAILS adds commands, MCP server and tool names and inputs; managed settings can lock the telemetry destination, and all operations in cloud sessions are logged for audit.

Sourcecode.claude.com/docs/en/monitoring-usage and docs/en/securityread 2026-09-29

Enterprise audit logs record actions taken by workspace members and which client each came from (Web, Desktop, iOS or Android), filterable by client or by member, with earlier entries marked Unknown. Dedicated pages cover project monitoring, logs, project usage, analytics and security insights. Build mode shows tasks as work proceeds, and project history records changes to a project. Audit logs are listed among Enterprise workspace capabilities, alongside security oversight across the workspace.

Sourcedocs.lovable.dev audit-logs, project-monitoring, logs, project-usage, analytics and security-insights pages, changelog and lovable-for-enterpriseread 2026-08-31

Claude Code runs on the developer's own machine by default, with code execution and file access staying local even in Remote Control sessions; cloud sessions run in isolated Anthropic managed VMs with configurable network access, or an organization routes them to self hosted environments on its own infrastructure, and dev containers add isolation. Data is encrypted in transit but not at rest, and Zero Data Retention is an enterprise arrangement rather than a default.

Sourcecode.claude.com/docs/en/securityread 2026-09-29

A default hosting region is set in Settings, under Security and compliance, Privacy and security, choosing Americas, Europe or Asia Pacific. On Business and Enterprise workspaces, admins can require a default region for every new project, after which members cannot choose a different one, and existing projects keep their original region.

Country regions including Germany, Japan and United States East are available on Enterprise by request, and Enterprise materials list regional code hosting. Lovable Cloud hosts apps with custom, verified and transferable domains, and two way GitHub sync and external deployment guidance let the app be hosted elsewhere.

Sourcedocs.lovable.dev privacy-and-security-settings, cloud, hosting, custom-domain, github and external-deployment-hosting pages, lovable-for-enterpriseread 2026-08-31

Plugins package skills, subagents, hooks and MCP servers as one installable unit, and Anthropic's official marketplace is added on first run and browsable from the /plugin Discover tab, with named plugins such as code-review, commit-commands, security guidance and skill-creator. Bundled skills (/code-review, /verify, /run, /batch, /simplify) and built in subagents ship with the product.

Sourcecode.claude.com/docs/en/plugins, docs/en/commands and docs/en/github-actionsread 2026-09-29

Business tier includes design templates, alongside a managed registry, a remix path for starting from an existing project, a prompting library, skills with reusable instructions, and design systems supplying customer components. Video tutorials and an idea to app guide cover getting started. There is no catalog of finished apps published by Lovable.

Sourcedocs.lovable.dev business/design-templates, managed-registry, projects/remix, prompting-library, skills and design-systems pagesread 2026-08-31

Every model Claude Code runs is Anthropic's. The customer picks among Claude models per session or subagent and picks where they are served (the Claude API, Amazon Bedrock, Google Cloud's Agent Platform or Microsoft Foundry), but not the model maker, and Anthropic does not support routing Claude Code to non-Claude models through any gateway.

Sourcecode.claude.com/docs/en/llm-gatewayread 2026-09-29

The builder has no model selection, model picker, bring your own key or provider setting, including in advanced settings, workspace admin settings and account settings. The model related integrations, Fireworks, Replicate, Perplexity and Gemini Enterprise, and an AI feature among the Cloud backend pages give model access to the app being built, not a choice over the model that powers Lovable.

Sourcedocs.lovable.dev sitemap inventory, advanced-settings, workspace-admin-settings, lovable-account-settings and features/ai pagesread 2026-08-31

The Claude Agent SDK gives the same tools, agent loop and context management that power Claude Code as a Python and TypeScript library, with hooks, subagents, MCP, permissions, sessions, skills and plugins, and the CLI can be driven from any language with -p and JSON output.

Sourcecode.claude.com/docs/en/agent-sdk/overviewread 2026-09-29

Lovable has its own API and MCP server. Agent integrations turn a published Lovable app into an MCP server that ChatGPT, Claude and other assistants can use, with configurable tools, sign in, permissions and access. Custom MCP servers, MCP registries, a create connector path and an any API route cover other services. Two way Git sync with GitHub and GitLab, including self managed and dedicated GitLab, exports the full codebase, and desktop, mobile and Slack surfaces extend access to the platform.

Sourcedocs.lovable.dev lovable-api, lovable-mcp-server, agent-integrations, custom-mcp, mcp-registries, create-connector, any-api and git-sync-overview pagesread 2026-08-31

claude plugin eval runs a suite of realistic prompts in fresh, isolated non interactive Claude Code sessions, grades each run with regex, tool call or model judged rubric graders on the reply, transcript or files, compares against a no plugin baseline, reports pass rates per case, and gates CI on the score, so a team can catch regressions when its setup or the model changes; /verify and /code-review check individual changes.

Sourcecode.claude.com/docs/en/plugin-evalsread 2026-09-29

Browser testing has the platform use the customer's app in a real browser running in a virtual environment, clicking buttons, filling forms, navigating pages and checking real user behavior with screenshots across screen sizes. It returns a verdict.

A Details view shows the steps taken and summarizes what worked and what did not, and the pattern is reproduce, fix, then verify. Where test and live environments are turned on, it runs against the project preview in the test environment, so the check happens before the change reaches anyone. The check is Lovable's own, run against the customer's app before release.

Sourcedocs.lovable.dev browser-testing page and test-and-live-environments documentationread 2026-08-31

The Chrome integration lets the agent operate a visible browser session with the user's logins, navigating, clicking, typing, filling forms, uploading files, reading console errors and DOM state and taking screenshots, in Chrome, Edge and other Chromium browsers, with site permissions set in the extension and a pause for logins and CAPTCHAs; computer use extends this to native macOS apps.

Sourcecode.claude.com/docs/en/chromeread 2026-09-29

Browser testing and the preview toolbar run and display the app the agent has built, which checks the agent's own output, not third party software. Firecrawl and Apify integrations retrieve web content over HTTP, which is scraping, not computer use. All outside actions run through connectors, the any API route, custom MCP servers or MCP registries, all programmatic interfaces. There is no browser navigation, form filling, screen control or computer use.

Sourcedocs.lovable.dev browser-testing, preview-toolbar, projects/preview, firecrawl and apify pages plus sitemap inventoryread 2026-08-31