Forall Review 2026: The Coding Agent That Proves Its Code Is Correct
Forall is a spec-driven coding agent from Astrio that generates code alongside machine-checkable proofs. We review how it works, its MCP verify-only mode, and who should use it.
Most coding agents will happily hand you code that looks right. Forall asks a harder question: can you prove it’s correct? It’s a coding agent from Astrio that generates spec-driven code alongside machine-checkable proofs — so the output isn’t just plausible, it’s verifiable.
What is Forall?
Forall is a coding agent that helps developers build correct software by generating code together with formal, machine-checkable proofs. You can run it two ways:
- Full CLI agent — specs, proofs, and the whole workflow live in your terminal.
- MCP verify-only — stay in Cursor, Claude Code, or GitHub Copilot and add Forall’s hosted verification as an MCP server. No CLI install required.
On first launch you sign in with a Forall account (API key) or bring your own model key (OpenAI / OpenRouter). Then forall init in a git repo and you’re working.
Key features
- Spec-driven code with proofs — the agent generates code and a machine-checkable argument that it meets the spec.
- Two deployment modes — full CLI, or MCP verify-only inside your existing agent.
- Language support — TypeScript, Java, and Rust today, with more on the way.
- Bring-your-own-model — OpenAI / OpenRouter, or a Forall account API key.
- Apache-2.0 and fully open source.
Who should use it?
Forall is built for developers and teams who can’t afford “plausible but wrong” — safety-critical code, financial systems, or anything where a silent bug is expensive. The proof layer is the point: instead of reviewing generated code by eye, you get a checkable claim that it satisfies the spec.
It’s less compelling if you just want the fastest possible codegen for a throwaway script — the proof step adds overhead you may not need there.
Pros and cons
Pros: verifiable correctness, flexible deployment (CLI or MCP), open source, BYO-model.
Cons: only three languages supported so far; proofs add workflow overhead; needs a Forall account or model key to start.
Pricing
Free and open source (Apache-2.0).
FAQ
Does Forall replace my current coding agent? Not necessarily. The MCP verify-only mode is designed to sit alongside Claude Code or Cursor and add proof-checking without moving your workflow.
Which languages are supported? TypeScript, Java, and Rust, with more planned based on demand.
Is it really free? Yes — the agent is Apache-2.0 open source. Running it still consumes tokens from whatever model you connect.
Explore the best AI Coding tools
Related Articles
agent-run Review 2026: Run Coding Agents in a Tiny Sandbox That Catches Mistakes Before They Spread
In-depth review of agent-run — a sub-1MB standalone binary that sandboxes coding agents (Claude Code, Codex, OpenCode, pi) inside a Bubblewrap container. Host filesystem is read-only by default. Built to catch agent mistakes, not malware.
Best AI Agent Tools in 2026: From Coding Assistants to Autonomous Workers
Complete guide to AI agent tools in 2026 — Claude Code, Codex, Cursor, Manus, and more. Which agents actually deliver on the promise of autonomous work?
Faultsense Review 2026: The expect() Without the Page
Faultsense is a zero-dependency browser agent that runs end-to-end assertions against real user sessions in production. We review how fs-* attributes work, RUM-style testing, and who should adopt it.
Best AI Coding Tools in 2026: Cursor vs GitHub Copilot
Compare the top 5 AI coding tools of 2026: Cursor, GitHub Copilot, Claude Code, v0, and Windsurf. Find the best AI pair programmer for your workflow.
Subscribe to the 9bests weekly — get the full list free
Hand-picked AI tool reviews and updates every week. Subscribe to receive this full list + 7 more quick-reference sheets (writing / image / video / audio / chat models / data / API cost).
Subscribe free & get it →Independent reviews — ratings aren't influenced by vendor payments · double opt-in · unsubscribe anytime