Coding agent · IDE (Fork)

Cursor

Verified Aug 2026

Quick answer: Daily-driver AI IDE: Composer 2.5 plus Grok 4.6, Grok Bot teammate on paid plans, tab completion, Design Mode, local/cloud agents, CLI/SDK, and an iOS companion to supervise runs.

Best for

Daily-driver AI IDE: Composer 2.5 plus Grok 4.6, Grok Bot teammate on paid plans, tab completion, Design Mode, local/cloud agents, CLI/SDK, and an iOS companion to supervise runs.

Skip if

You need a free always-on agent loop, strict local-only BYOK with no vendor models, a lightweight VS Code plugin instead of a full fork, or a GitHub replacement you can treat as source of truth today (Origin is early-beta hosting, not that).

Who it fits

  • Daily-driver AI IDE: Composer 2.5 plus Grok 4.6, Grok Bot teammate on paid plans, tab completion, Design Mode, local/cloud agents, CLI/SDK, and an iOS companion to supervise runs.
  • People who prefer a dedicated desktop coding surface
  • Terminal-first engineers who live in the shell and git
  • Builders who start in the browser without a local IDE setup
  • Users who supervise or kick off agent runs from mobile
Editor's Choice50% off with the link below

Pros

In-house Cursor Models — Composer 2.5 (~10× cheaper than Opus for daily coding) plus Grok 4.6 (joint with SpaceXAI) for long-running in-IDE agents; Grok Bot always-on teammate (desktop/mobile, cloud computer, routines — separate weekly usage pool) included on Pro, Pro+, Ultra, and self-serve Teams; blazing tab completion with million-token context; Design Mode for quick frontend edits; local and cloud agent orchestration with automations and loops; CLI and SDK; official iOS app to start and supervise cloud agents on the go.

Cons

Can be resource-heavy; monthly limits can be hit quickly when using external models via API; Grok Bot draws from its own weekly Cursor-metered pool (not Composer/Grok 4.6 model usage); Enterprise Grok Bot requires admin enablement; subagent orchestration is a bit less turnkey than dedicated orchestrators; iOS companion supervises agents — not a full mobile IDE.

Origin (git forge)

Cursor Origin is Cursor’s git forge — code hosting, pull requests, browse/search, and an origin CLI — not a tab inside the editor. It began rolling out 17 August 2026 as early beta on paid Cursor plans (Pro, Teams, Enterprise). Free accounts cannot store code there.

I mirrored GitHub repositories into Origin and did not switch remotes. That is the product’s own adoption path: GitHub stays the source of truth; Origin copies history, branches, tags, and pull requests; comments sync both ways within seconds; pushes on a mirrored repo still land on GitHub. Detach from GitHub only if you want Origin to own the remote.

Feature parity is not a reason to move. Origin does not bring GitHub Issues, GitHub Actions, or the marketplace. Depot and Buildkite can run existing Actions YAML, but only on Origin-hosted repos — not on GitHub mirrors. Cursor has not published retention, residency, or training-use terms for Origin; the host now sits under SpaceX / SpaceXAI. Treat a mirror as a convenience window until those terms exist and the beta has been through a few GitHub outage weeks.

The loop that actually gets closer to Cursor’s infra is cloud agents on a native Origin repo. Those VMs clone into Cursor’s fleet; GitHub HTTPS is no longer the middle of every clone/push, which matters when many agents hit one repo. Cloud agents already clone GitHub today — Origin is not required. Local Cursor agents already have the working tree on disk, so Origin vs GitHub does nothing for them. GitHub Actions minute caps and cancel-queues are a runner-billing problem; Origin does not fix them while the repo stays mirrored.

Cursor’s published Origin roadmap is one sentence: agent-native features ship soon. The June 2026 Compile pitch (Graphite stacked PRs, merge queues, machine-readable review for parallel agents) is the differentiator they are aiming at — not “we also have a PR tab.” Revisit a remote switch when that layer ships and Origin looks like boring infrastructure, not a research preview.

Origin docsOrigin changelogcursor.com/codebase

Grok Bot (teammate app)

Grok Bot is Cursor’s always-on teammate — desktop and mobile apps with its own cloud computer, routines, and a weekly usage pool metered on your Cursor account. It is not Grok 4.6 inside the editor; Composer 2.5 and Grok 4.6 remain the in-IDE Cursor Models for coding agents.

Access is included on every paid individual Cursor plan (Pro, Pro+, Ultra) and on self-serve Cursor Teams for all members — no separate Grok Bot subscription and no Premium seat required. Weekly included usage resets; extra work can continue on shared on-demand spend if enabled. Enterprise teams must enable access through their account executive.

A free trial is a usage credit (with a 7-day window), not a paid SKU. Individual SuperGrok, SuperGrok Plus, or SuperGrok Heavy accounts can link for additional Grok Bot usage grants; SuperGrok Lite and team/enterprise SuperGrok tiers do not link.

Grok Bot plans

Facts

Locality:
Hybrid (local + cloud)
Surfaces:
Desktop, CLI, Web, Mobile
Maturity:
Preview / beta
Free tier:
Free option available
Cloud agents:
Remote agent execution
Verified:
Aug 2026

Agent standards & memory

AGENTS.md: Full SupportSKILL.md: SupportedRules / memory: .cursor/rules (Project, User, and Team scopes); Persistent memory.Basic free tier (limited premium requests). Grok Bot included on paid individual plans and Teams (weekly allowance; trial is usage credit, not a paid plan).

Mobile

App Store

Compare one neighbor first

Vendor pages sell hard. Read one alternative on this site, then leave for the official link if the fit still holds.

Alternatives

Related tools

Someone asks you about vibe-coding software? Share this site with them — we are adding more useful data regularly. agents.dancingteeth.net

← Back to full tool directory