Anthropic Claude Code vs Factory (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, claude.ai connectors, and managed settings that allow or deny servers across an organization, alongside built in file, search, shell and web tools.

Sourcecode.claude.com/docs/en/mcp and docs/en/chromeread 2026-10-24

MCP connects Droid to the customer's own tools and data, and Connectors add third party apps with managed authentication and organization level controls. Delegations take work from Slack, Linear, Jira, Microsoft Teams and Google Chat with shared instructions, plugins, identity and compute, and automations act on GitHub events and inbound webhooks, with org policy deciding which models, tools and integrations Droid may use.

Sourcedocs.factory.com/harness/mcp, harness/connectors, delegations and enterprise/identity-and-access; docs.factory.com/delegations/indexread 2026-10-28

Subagents are workers Claude spawns for delegated tasks, and agent teams let a lead agent supervise long running peer sessions that share a task list. Dynamic workflows go further. Claude writes a JavaScript script that runs dozens to hundreds of subagents in the background, up to 1,000 per run and 16 at a time by default, asks before a run starts (every run in Manual mode, the first launch in Auto mode), and the script can be saved and rerun as a command.

Sourcecode.claude.com/docs/en/workflowsread 2026-10-24

Custom droids are subagents with their own system prompt, model and tool policy that Droid delegates focused tasks to in a fresh context window. Missions plan and run large multi feature projects: the customer approves a plan of features and milestones, and an orchestrator in Mission Control assigns the work to workers, validates milestones and runs headless through droid exec --mission. Worktrees keep parallel sessions apart.

Sourcedocs.factory.com/harness/subagents, missions/overview and missions/referenceread 2026-10-28

Cloud scheduled tasks, in research preview, run a saved prompt against chosen repositories on a schedule, on an HTTP POST to the task's own endpoint, or on GitHub pull request and release events, so they keep working with the laptop closed. GitHub Actions and GitLab CI/CD automate code review and issue triage in CI, GitHub Code Review reviews every pull request automatically, and the agent answers @Claude in Slack and runs from the terminal, desktop, IDEs, the web and headless scripts.

Sourcecode.claude.com/docs/en/routines and docs/en/overviewread 2026-10-24

Automations start a Droid session on a schedule (plain words or cron), a Slack message, a GitHub event or an inbound webhook, running on a Droid Computer, an Environment, the desktop app or GitHub Actions. Incident Response sets Droid to respond to production alerts in Slack, Triage classifies inbound work and routes it, and Droid Exec publishes GitHub Actions workflows for PR review and weekly cron runs.

Sourcedocs.factory.com/software-factory/automations, incident-response, triage and droid-exec/overviewread 2026-10-28

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 2026-10-24

AutoWiki generates structured wikis of each repository's architecture, modules, APIs and conventions. The /wiki command builds one from the Droid CLI, /install-wiki sets up a CI action that regenerates it on every push to the default branch, the Wiki Browser in the Factory App searches and manages wikis, and the AutoWiki API creates runs, browses pages, searches and exports.

Sourcedocs.factory.com/software-factory/wiki/overview, wiki/auto-refresh, wiki/web-viewer and api-reference/wikiread 2026-10-28

Auto memory, on by default in local sessions, holds notes Claude writes itself as it works, such as corrections the user gives and approaches the user confirms. It is scoped per repository and shared across worktrees, 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 2026-10-24

Sessions are resumable across the app, CLI, web and mobile and drivable through the Sessions API, and AGENTS.md and hierarchical settings carry configuration across sessions. Session state stays within one session, AGENTS.md holds standing instructions, and the documentation index lists no memory layer of Factory's own.

Sourcedocs.factory.com/api-reference/sessions, harness/agents-md and llms.txtread 2026-10-28

Auto mode is the starting mode for interactive terminal and VS Code sessions from v2.1.283. A separate classifier model reviews each action before it runs and blocks what it judges unsafe, the user's ask and deny rules still apply, and an organization can turn auto mode off in managed settings. Manual mode starts read only and asks before edits, shell commands and network requests, with approve once or allow from then on and fail closed matching for unmatched commands. Plan mode researches and proposes changes without editing, and claude -p and Agent SDK runs start in Manual.

Sourcecode.claude.com/docs/en/permission-modes and docs/en/securityread 2026-10-24

Autonomy levels (Off, Low, Medium, High) decide what Droid may do without confirmation, and permission rules allow, require approval for, or block matching shell commands, with an ask rule requiring approval even at High autonomy. Missions start only after the customer approves the plan, org managed settings enforce these as hard policy, and OS level sandboxing and Droid Shield secret scanning sit beside them.

Sourcedocs.factory.com/autonomy-and-safety/permission-rules, auto-run and missions/planningread 2026-10-28

A SOC 2 Type 2 report and an ISO 27001 certificate sit in the Anthropic Trust Center. Claude Code adds permission modes, sandboxed shell commands with filesystem and network isolation, command injection detection, trust verification for new folders and MCP servers, keychain credential storage, managed settings for organization enforcement and a HackerOne disclosure program. Under commercial terms Anthropic does not train models on Claude Code code or prompts unless the customer opts in.

Sourcecode.claude.com/docs/en/security and docs/en/data-usageread 2026-10-24

Factory lists SOC 2 Type II, ISO 27001 and ISO 42001 with reports, architecture documents and subprocessor lists in its Trust Center at trust.factory.com, and offers SSO over SAML or OIDC, SCIM directory sync, organization roles flowing from the identity provider, service accounts with their own API keys and an organization audit log that records authentication, SSO and SCIM changes and policy changes.

Sourcedocs.factory.com/enterprise/compliance-audit-and-monitoring and enterprise/identity-and-accessread 2026-10-28

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. Telemetry endpoints set in managed settings override developer settings, and all operations in cloud sessions are logged for audit.

Sourcecode.claude.com/docs/en/monitoring-usage and docs/en/securityread 2026-10-24

Droid exports OpenTelemetry to the customer's collector: each user turn produces a trace whose child spans are the model calls and tool runs, with droid.tool and droid.message spans when message content logging is on, plus tool, MCP and blocked command metrics tagged by session and user. An organization audit log records who did what, including who started a Droid session from a connected tool, and Factory Analytics reports usage, credits and tool use per session.

Sourcedocs.factory.com/enterprise/telemetry/data-reference, audit-log and compliance-audit-and-monitoringread 2026-10-28

Claude Code runs on the developer's own machine by default, and Remote Control keeps code execution and file access local. Cloud sessions run in isolated Anthropic managed VMs with network access limited by default, or an organization routes them, scheduled tasks included, to self hosted environments on its own infrastructure.

Prompts and outputs travel over TLS 1.2 or later, and at rest the Anthropic API uses AES-256 disk encryption, while Amazon Bedrock, Google Cloud's Agent Platform and Microsoft Foundry apply their own keys, with customer managed keys available on the first two. Commercial data is kept 30 days by default, and Zero Data Retention is turned on per organization for qualified Claude for Enterprise accounts.

Sourcecode.claude.com/docs/en/data-usage and docs/en/securityread 2026-10-24

Factory offers two deployment options. Factory Managed runs the control plane in Factory's cloud while Droid runs on developer machines, CI, machines the customer registers through Remote Control and Factory managed Droid Computers, with LLM traffic routed to the customer's own providers or gateways and a dedicated EU deployment (api.eu.factory.ai, EU session storage and EU only inference) for enterprise customers.

Factory Private, in private preview, runs Droid sandboxes inside the customer's own VPC through OpenCompute, a self hosted service that provisions sandboxes on Kubernetes, EC2 or Lambda microVMs on demand. A fully airgapped Custom Deployment uses in network model endpoints with no Factory cloud at runtime, and Enterprise adds data residency, dedicated compute and on premises options.

Sourcedocs.factory.com/enterprise/network-and-deployment, opencompute/overview, enterprise/airgapped-deployment and factory.com/pricingread 2026-10-28

Plugins package skills, agents, hooks and MCP servers as one installable unit, and Anthropic's official marketplace is added the first time an interactive terminal session starts, browsable from the /plugin Discover tab or Claude Marketplace on the web, with a context cost estimate for each official plugin and named plugins such as commit-commands. Saved workflows and the bundled /deep-research workflow run as commands, and workflows can ship inside plugins.

Sourcecode.claude.com/docs/en/plugins and docs/en/workflowsread 2026-10-24

Plugins package skills, commands, custom droids, output styles, hooks and MCP servers, and Factory maintains an official marketplace at Factory-AI/factory-plugins naming droid-control (terminal, browser and computer automation for QA and demos), droid-evolved (skills for navigation, writing, design and frontend work) and security-engineer (security review, threat modeling and vulnerability validation). Organizations run internal approved marketplaces, and since 27 October admins pick plugins from their registered marketplaces.

Sourcedocs.factory.com/harness/plugins, enterprise/internal-plugin-marketplaces and changelog/release-notesread 2026-10-28

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 2026-10-24

The customer picks among Factory hosted models from Anthropic, OpenAI, Google, xAI and Factory's own pool, lets the Factory Router choose per task, or brings its own OpenAI, Anthropic, open source or local models by key, and enterprise controls restrict which models, base URLs and providers a fleet may use.

Sourcedocs.factory.com/models and model-independence/byokread 2026-10-28

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 2026-10-24

A REST API at api.factory.ai with bearer API keys creates and drives Droid sessions, wiki runs, CI automations and readiness reports (the Sessions API is enabled for selected organizations), beside the Droid SDK and headless droid exec for scripts and CI.

Sourcedocs.factory.ai/api-reference/sessions and droid-exec/overviewread 2026-09-29

claude plugin eval runs a suite of realistic prompts against a plugin and scores the results, grading each case with a regex over the reply, a check that a particular tool was called, or a rubric a second model judges. It compares the run against a no plugin baseline and gates CI on the score, so a team can catch regressions when its plugin changes or a new model ships.

Sourcecode.claude.com/docs/en/plugin-evalsread 2026-10-24

The Agent Readiness model scores how ready the customer's repository is for agents, Agent Effectiveness measures throughput, output and spend across the organization, and code review, security review and Automated QA test the customer's code. Factory publishes its own benchmark results (Agent Arena, Legacy Bench, Terminal Bench). Factory does not describe a harness, scored test set or optimization loop for evaluating Droid's behavior on the customer's work.

Sourcedocs.factory.com/agent-readiness/overview, agent-effectiveness/overview and benchmarksread 2026-10-28

The Chrome integration lets the agent work in a visible browser window with the user's logins, navigating, clicking, typing, filling forms, uploading files, reading console errors and DOM state and saving screenshots, in Chrome, Edge and other Chromium browsers. Site permissions come from the extension, and the agent pauses for logins and CAPTCHAs. Computer use extends this to native macOS apps.

Sourcecode.claude.com/docs/en/chromeread 2026-10-24

Droid Control, an official Factory plugin, lets Droids launch and operate web apps and Electron apps through end to end flows. The Droid clicks, types and captures screenshots as QA and demo evidence, and Automated QA tests the application as a real user would with visual evidence. The CLI also ships an agent-browser skill for browser automation, the browser pane in Factory's desktop app takes Droid's clicks and keystrokes directly, and since 20 October the built in browser can import Chrome cookies on macOS so Droid stays signed in.

Sourcedocs.factory.com/software-factory/droid-control, automated-qa and changelog/release-notesread 2026-10-28