Developer tools decision
Cursor
A narrow single-user coding-agent that reads a repo, calls off-the-shelf LLMs, and applies patches is realistic for a capable engineer to build and maintain; Cursor’s proprietary models, enterprise features, hosted cloud agents, and compliance claims make a full replacement impractical for small teams.
Visit website↗Open-source builds that already do this
Every project below is open source and already does this job today. Fork one, self-host it, or take the parts you need - the build prompt further down assumes an empty file, and this is the shortcut past that. Licences differ; check the one on each card before you ship. All Cursor alternatives, with the arithmetic →
What a replacement has to do
- User points Cursor at a repository; agent ingests repo context, calls an LLM, generates code edits/patches, runs tests/commands, and applies changes.
What it still won’t have
- Cursor’s proprietary model(s) (Grok 4.5 and Cursor-hosted models)
- Built-in enterprise features (pooled usage, SSO/SCIM, audit logs, invoicing)
- Hosted cloud agents, marketplace plugins, and native desktop integrations
- SOC 2 / vendor compliance guarantees and vendor-provided privacy controls
What remains hard
- Proprietary models
Our most intelligent model and the first we've built for more than software engineering.
- Brand trust
Trusted by over half of the Fortune 500 to accelerate development, securely and at scale.
First-year cost
Keep paying
Paying is—cheaper in year one.
On cash alone, building overtakes the subscription at 6 seats.
Money you would actually spend
Time you would spend
—
What you would spend
What we assumed
The verdict above measures whether you could build it. This one is only about money.
Runnable build prompt
Build a single-user AI coding agent CLI + minimal web UI using Node.js (TypeScript), a small React front-end, and a Postgres or SQLite store. Core features in scope: (1) repo indexing (walk files, build token-aware context selector), (2) prompt assembly and LLM integration (OpenAI/Anthropic API support), (3) apply/preview patch/diff workflow with git integration, (4) run selected shell commands/tests in a sandbox and capture output, (5) basic subscription/auth (local user account) and usage logging. Out of scope: multi-user team billing, SSO/SCIM, enterprise audit logs, training proprietary models, cloud agent fleet orchestration, and SOC2 compliance. Include robust error handling, rate-limit/backoff for API calls, and unit tests for prompt assembly, patch application, and sandbox execution.
How we checked
How the score was reached
- Partly verdict base52
- An open-source build was found+5
- 5 cited sources+3
- Price verified on pricing page+3
- 3/3 assessment runs agreed+4
- Hard moats found in the evidence-3
- Evidence score64
The base comes from the verdict. Everything under it is a check that either happened or did not, and each one is a fact frozen in this record rather than a judgement made at render time - so the same evidence always produces the same number.
How scoring works →Cited sources · 5
Every page the run actually retrieved.
- official productCursor — AI coding agent
- official pricingCursor Pricing
- official docsCursor — Build Software with AI Agents
- open sourceTabbyML/tabby
- open sourceiOfficeAI/AionUi
Integrity checks
What held up, and what did not.
