build edition · migrate

No migration.
Just an extra layer.

fastpace doesn't replace your AI agent. It stacks on top. Keep Cursor, Copilot, Cline, Claude Code — whatever you pay for — and add domain context + spec discipline + signed receipts. Same agent, dramatically better output.

See pricing How it works →
Stack on top of your current AI agent

One install. Six setups. ~5 minutes each.

Click the agent you currently use. Concrete setup. Honest cost-stack math. What you actually get back.

Cursor

$20/dev/mo

fastpace adds domain context (L1–L5), spec-driven workflow (PRD → ERD → plan), and signed receipts on every commit. Cursor types; fastpace makes the typing fit your codebase + your backlog.

Setup:
  1. npm install -g @fastpace-ai/fp
  2. fastpace init in your repo. Writes fastpace/ alongside your .cursor/.
  3. Cursor reads your .cursorrules + the new fastpace/context/ automatically — no Cursor config change.
  4. /fp-discover to seed L1–L2 context (architecture, patterns, glossary).
  5. /fp-write-prd on your next feature instead of "build the X feature."
Stacked cost: $20 + $29 = $49/dev/mo
Net gain: AI output that ships first-time. Reviewer cycles drop ~40% on AI-generated PRs (per design-partner data).

GitHub Copilot Business

$19/dev/mo

Copilot has no domain context, no spec workflow, no Jira sync. fastpace adds all three. Copilot keeps autocomplete; fastpace handles the heavy lifting (PRD, ERD, plan, review, commit provenance).

Setup:
  1. npm install -g @fastpace-ai/fp
  2. fastpace init in your repo.
  3. Copilot keeps working as-is. fastpace runs in your terminal as fp for the spec workflow + your AI runtime of choice (Claude Code, etc.) for the actual generation.
  4. fastpace integrations setup jira (or linear) so the agent pulls real backlog items.
Stacked cost: $19 + $29 = $48/dev/mo
Net gain: Copilot autocompletes are still fast. fastpace is the layer that makes "build the X feature" turn into a shippable PR instead of a draft.

Claude Code (Anthropic CLI)

$$ usage-based

Native pairing. fastpace was built around Claude Code first. Skills, agents, hooks all install via <code>fastpace init</code> directly into <code>.claude/</code>. The recommended setup.

Setup:
  1. npm install -g @fastpace-ai/fp
  2. fastpace init — installs all 25+ skills, 7 agents, 16 hooks into .claude/.
  3. claude picks up the new skills automatically.
  4. /fp-discover in Claude Code seeds context.
Stacked cost: Claude Code usage + $29/dev/mo
Net gain: The intended pairing. Both tools stack cleanly.

Cline / Continue (open-source)

Free (BYO API key)

Drop-in. fastpace context loads automatically; hooks fire on every tool call. Lower-friction pairing because Cline/Continue are extension-based.

Setup:
  1. npm install -g @fastpace-ai/fp
  2. fastpace init in your repo.
  3. Cline/Continue read fastpace/context/ via the standard rules-file protocol.
  4. Hooks fire on every tool invocation — they handle the auditability layer transparently.
Stacked cost: Just $29/dev/mo for Team tier (Cline/Continue stay free).
Net gain: Cheapest path to spec-driven AI coding. Free agent runtime + $29 for the coordination layer.

Codex CLI

$$ usage-based

Hooks + provenance trailers. Run manifests captured per call. Same model as Claude Code, smaller surface.

Setup:
  1. npm install -g @fastpace-ai/fp
  2. fastpace init.
  3. Codex picks up the hooks automatically. Spec workflow runs through your preferred runtime.
Stacked cost: Codex usage + $29/dev/mo
Net gain: Codex velocity + fastpace shipping discipline.

Bedrock / Azure OpenAI / Vertex AI

Cloud-provider rates

Runtime-agnostic. Same fastpace governance pipeline regardless of which cloud you call. Your enterprise IT keeps its existing AI provider relationship.

Setup:
  1. npm install -g @fastpace-ai/fp
  2. Configure your runtime endpoint via fastpace runtimes add.
  3. fastpace mediates every call — same hooks, manifests, provenance whether the underlying model is Anthropic on Bedrock, GPT on Azure, or Gemini on Vertex.
Stacked cost: Cloud bill + $29/dev/mo (or Org tier for centrally-managed deployments).
Net gain: Multi-cloud shops get one governance surface instead of three.
FAQ

Common questions about stacking

Do I have to drop Cursor / Copilot / Cline?
No. Keep them. fastpace is the layer above your AI agent — it feeds the agent your domain context, your PRDs, and your linked backlog ticket so the output is shippable. The agent stays as-is.
Why pay for two tools?
Because they do different jobs. Cursor / Copilot are autocomplete + chat. fastpace is the spec discipline + context + traceability. Stacked, you stop rewriting AI output. The math: $20 (Cursor) + $29 (fastpace Team) = $49/dev/mo for code your reviewers approve first time. The savings show up in dropped review cycles, not in the tool budget.
How long does the migration take?
There is no migration. npm install -g @fastpace-ai/fp, then fastpace init in your repo. ~5 minutes. Your existing AI agent keeps working without any config change.
What if my team is already on a homegrown rules.md / .cursorrules?
fastpace reads those automatically as L2 context (architecture/patterns layer). You don't lose your existing investment — you build on top of it. The L3 (PRDs/ERDs) and L4 (execution plan) layers are net-new value.
Will fastpace add latency to my AI agent?
Hooks run inline (PreToolUse + PostToolUse). The latency budget per hook is tracked + benchmarked — see fastpace bench. Median impact is <50ms per AI call. Skill workflows are explicit slash commands, so no implicit overhead.
Does this work with my custom internal AI gateway?
Yes. The runtime registry (fastpace runtimes) is an allow-list of model endpoints — point it at your internal gateway and the rest of the pipeline works transparently. No model-specific assumptions.

Five minutes, one install, no migration.

Free for solo devs. $29/seat/mo for teams. Works with the AI agent you already pay for.

See pricing View on GitHub