Forge is a plugin that brings your engineering process into your AI coding tools. It guides work through planning, testing, and review, using your requirements, architecture, and team standards. Your AI tool does the coding; Forge supplies the workflow steps, configured approval points, and a record of the results. With Jira, Linear, or Notion connected, results are written to the ticket.
Your rules guide the code. Forge guides the steps. Your AI tools already support instructions and skills. Forge brings those into a repeatable workflow: brief the AI on the work, run the relevant checks, pause at configured approval points, and record the results. Teams can share that process across tools and engineers.
No. Forge runs inside Claude Code, Codex, and Cursor — the tools they already use. No new app, no retraining. An engineer kicks off a workflow (with the plugin, “implement PROJ-784” is enough), at the start of a session or partway through one, and keeps working in their tool while Forge handles the process around them. The skills and conventions your team already wrote become steps Forge can use.
No — you can pull Forge in at any point. Say “forge, fix this” or “forge, review the PR for PROJ-412” forty turns in, and the workflow picks up from where you are: the conversation, the files you opened, and the decisions you made carry into it.
Admins can also switch on session observation, and Forge will spot product work running outside a workflow on its own and offer to pick it up — no one has to remember. See how mid-session pickup works →
It's built to do the opposite. Forge briefs the AI with your codebase and standards up front, so engineers stop re-explaining context and prompt-wrangling every session — and the first pass holds up instead of bouncing back in review. It only pauses where you've decided a human should sign off; everything else runs automatically. Less busywork, not more process.
No. Your existing instructions stay exactly as they are. CLAUDE.md, .cursorrules, and other tool configs keep governing how the AI writes code in a session. Forge operates at the layer above: what the AI is scoped to work on, what context it gets, which checkpoints it must pass, and what gets logged. Your coding conventions are never touched.
Yes. Every workflow is a list of steps you control — each with a model tier, an approval gate, and any required skills. An admin can edit, reorder, or disable any step, set it once for the whole org, or fork it for a single team. You can even describe a new workflow in plain English from inside your AI tool, and Forge shows exactly what changes before it saves.
Every workflow step has a target model tier—fast, balanced, or capable. Routine steps like fetching a ticket or posting status run on the cheaper, faster models; only the hard, code-heavy steps reach the most capable one. You get the right result at the right price without anyone switching models by hand.
The tier is a target matched in both directions: a session on a frontier model hands routine steps down to a cheaper one, and a session on a cheaper model escalates the hard steps up. The hand-off runs in an isolated sub-agent with a clean context for that one step, so quality is never traded for cost.
Model delegation is supported in Claude Code and Codex. Availability depends on your AI tool; without delegation or model switching, the workflow uses the model already running your session.
It reads the usage your tools already record — input, output, and cache tokens — with no proxy between you and the model. It normalizes every model and tool to a common “weighted token” so spend is comparable on one axis, then prices it at published list rates. It's a transparent estimate, and it says so: it excludes flat-rate subscription plans and any negotiated discounts.
AI tools: Claude Code, Codex, and Cursor each have a dedicated plugin, plus anything that supports MCP, the open Model Context Protocol. The plugins add automatic routing and guardrails. Token measurement is available in Claude Code and supported Codex versions; Cursor usage is currently unmeasured.
Jira, Linear, and Notion connect through your AI tool's own integrations, so Forge writes results to the ticket you already use. You can start without one: workflows use the context in your AI tool, and results that would go to the ticket are handed to you to paste.
Create a free account, install the plugin in your AI tool, and run guided setup. Setup recommends a starter set of workflows for you or an admin to approve. Then try one real task. Individual is free forever. Team includes shared workflows, roles, organization-wide analytics, and Token Intelligence for up to 10 members. Team is free during beta, then $50/month. Business is coming soon.
Your AI tool reads and changes code using its existing access. Forge supplies workflow instructions; its service does not directly access your repository, CI/CD, or production. It never receives your API keys, passwords, or tokens (your tool handles those), and it never trains any model on your data. Forge receives structured workflow metadata, including steps, timing, token usage, and work-item keys, titles, and links. It does not receive your source code or raw conversations. Full detail is on our Privacy page.