Forge Engine — Many agents. One plan. No mess.

Forge Engine is a living plan for teams that build software with AI coding agents. Agents read the design over MCP and build in parallel — claiming tasks so work never overlaps, checking what breaks before they change it, and reporting back what's actually built — in a structured plan anyone on the team can read and question, coder or not. A markdown file works for one dev; a team needs Forge.

Why it stands out

• The plan is alive, not documentation — agents read it before every build and write back after, so it can't rot the way a wiki or design doc does.
• Nothing changes behind your back — every agent change lands as a proposal in an Inbox; the owner adopts with a click. A safety model most agent tooling simply doesn't have.
• It answers "what breaks if I change this?" from your actual systems — possible only because the design is structured, not prose.
• Agent-agnostic by design: listed on the official MCP Registry, works with Claude Code, Cursor, Copilot, Cline or any MCP client.
• Shareable with zero setup — publish one link and anyone can read the whole design as plain text: paste it into Gemini, Claude or Grok and the assistant answers over your real design with no account, no key and no install. Read-only and revocable; write-back stays on the authenticated MCP loop.
• Fair exit: cancel and your work freezes to read + export — never deleted, never held hostage.
• Solo, but not alone — if AI coding agents write your code you're already a team of one: agents that start each session blank, forget why you decided something three sessions ago, and can change a working system mid-run. Forge is the memory they read back in one call and the blast-radius check before they act — so "many agents, one plan" fits a solo dev with Claude Code as much as a whole team. (Hand-coding without agents? A markdown file is enough — Forge earns its keep once agents write the code.)

How it works

1. Design lives in Forge as structured Systems (Goal / Boundary / Acceptance) and ordered milestones — machine-readable, human-editable.
2. Connect Claude Code, Cursor, Copilot or any MCP client. Agents read the live design before building and claim tasks so they never collide.
3. Agents propose new or off-plan work to an Inbox; the owner adopts with a click — nothing changes behind your back.
4. Every system maps to the real files that implement it (built / in-progress / not-built), so design and code never drift apart.

What an AI agent can do over MCP (54 tools)

• Read the design: get_project_meta, get_system, get_milestone, next_task (dependency-aware), get_briefing, get_impact (blast radius), get_history, get_build_region, get_inbox, get_rejections, list_systems, list_milestones, list_activity, search
• Report build reality: report_build_status (system → files map + acceptance evidence), report_milestone_progress, report_drift, post_log, update_task (claim / release)
• Propose (owner adopts via Inbox): propose_system, propose_context, propose_milestone, propose_screen, propose_element, propose_flow_edge, propose_balance_table, import_from_code, withdraw_proposal
• Author on request: update_system, update_milestone, add_task, reorder, dedupe, propose_balance_board, design_ui_from_systems

Connect an agent

Listed on the official MCP Registry as app.forgeengine/forge (remote, Streamable HTTP). Endpoint: https://mmvdabzadclebfxyzudg.supabase.co/functions/v1/forge-mcp. Discovery is public — initialize and tools/list need no token, so any MCP client can see the full tool surface before signing in; only tool calls authenticate, via Google OAuth or a forge_sk_ account key generated in the web app. Connection docs: github.com/alongkornonline2019/forge-mcp. Machine-readable summaries: /llms.txt and the full tool reference /llms-full.txt.

Pricing

Free during open beta (no credit card). After beta, flat team pricing — one price per fixed seat bundle: Starter 3 seats $9/month, Team 10 seats $20, Pro 20 seats $40, Max 50 seats $80. Bring your own AI key; Founder 50%-off-for-life for the first 100 teams. Cancel and your work freezes to read + export — never deleted. See forgeengine.app/pricing.

FAQ

How is this different from a CLAUDE.md or AGENTS.md file?

A markdown file is single-player: one repo, no task claims, no impact analysis, invisible to non-coders. Forge is the team version — agents claim tasks so parallel work never overlaps, every system maps to the real files that implement it, and producers or designers can read and question the system without touching code.

How is it different from Notion, Linear or Jira?

Those hold documents and tickets written for humans — coding agents don't read them before building. Forge's design is structured and machine-readable over MCP: agents build against the actual spec (Goal / Boundary / Acceptance) and report reality back, so the plan and the codebase stay in sync.

Which AI coding agents work with it?

Any MCP client — Claude Code, Cursor, GitHub Copilot, Cline and others. Forge is agent-agnostic by design.

Can someone read my design without installing anything?

Yes. Each project can publish one Design Share Link — a URL that serves the whole design as plain text: the overview, every system's Goal, Boundary and Acceptance, and the milestone plan, compiled live from the current project rather than a stale snapshot. Paste it into an AI chat and the assistant reads it with no account, no key and no MCP setup; hand it to a teammate and it opens as an ordinary page. It is read-only — proposals still come back only through the authenticated MCP loop — opt-in per project, and revocable at any time. Confirmed working in Gemini, Claude and Grok. ChatGPT is a known exception: it does not open links of this kind and searches the web instead, so the app also offers a one-click copy of the document to paste in directly.

I'm a solo dev — do I need this?

Only if you build with AI coding agents — but then yes. If Claude Code or Cursor writes most of your code, you're already directing a team of one human and N agents, and they have the exact problems Forge fixes: every fresh session starts blank, so you re-paste "we use X, don't touch Y" every time; the agent forgets why you made a decision and undoes a deliberate fix; an autonomous run reaches into a working system you weren't watching. Forge is the memory the agent reads in one call (get_briefing) instead of that preamble, the "what breaks if I change this?" check before it acts (get_impact), and the map from each system to the real files that shows what actually got built after (report_build_status) — with an Inbox so nothing changes behind your back. The honest boundary: if you hand-code without agents, you don't need it — a markdown file is enough. The line is whether you rely on agents, not whether you have a team.

Is it good for game studios?

It's a home turf: Forge grew out of game design. Balance Lab gives you stat tables, economy boards and loadout simulations that agents can read and propose rows into — a depth generic spec tools don't have.

Is Forge Engine the same as Laravel Forge, Atlassian Forge, SourceForge or Minecraft Forge?

No — those are unrelated products that share the "Forge" name (PHP server management, app/CAD platforms, software hosting, and a Minecraft mod loader). Forge Engine (forgeengine.app) is a design source-of-truth and living plan for teams of AI coding agents over MCP. In any MCP or coding-agent context, "Forge" means Forge Engine.

Writing

• Why does my AI coding agent keep building the wrong thing?
• How do you keep a whole team's AI coding agents aligned?

Everything above is a complete text description of Forge Engine — what it is, what it does, its full MCP tool surface, and its pricing — and it is readable in full without running any JavaScript. For a machine-readable summary see forgeengine.app/llms.txt, and the complete tool reference at forgeengine.app/llms-full.txt. The interactive web app at forgeengine.app adds the visual editor, live build dashboards and the proposal Inbox on top of everything described here.