Spiral 4.0 adds Style Engine stylometry and MCP support
Spiral 4.0 added a stylometry-based Style Engine for matching a writer or brand voice and shipped MCP plus CLI access for agent workflows. It also now plugs into Claude Code, Codex, and OpenClaw, and the base price drops to $15.

TL;DR
- Spiral 4.0 adds a new stylometry-based Style Engine that, according to danshipper's launch post and Every's official launch write-up, is meant to learn a writer or brand voice from past samples and reuse it in new drafts.
- danshipper's launch post says Spiral is now built for agent workflows, and the company's agent setup page shows remote MCP access plus OAuth sign-in for clients like Claude Code.
- The same launch post and the linked Spiral site position the product as a writing harness that can be called from CLI tools such as Codex, Claude Code, and OpenClaw.
- A retweeted pricing post says the base price drops from $25 to $15, while Every's official launch post says the product is also moving from session limits to token-based billing.
You can read Every's full launch post, browse Spiral's agent setup doc, and check the live product page at writewithspiral.com. The sharpest detail in the release is the stylometry claim, Every says the new engine produces drafts that blend in with a user's real samples 87 percent of the time. The other notable shift is distribution: Spiral is no longer framed as a browser writing app first, it is being pushed into agent clients and command-line workflows.
Style Engine
The headline feature is a rebuilt Style Engine based on stylometry, which danshipper's launch post describes as a way to extract a writer or brand voice from prior work. Every's official launch post adds the technical framing: Spiral computes a writing fingerprint, selects relevant samples for a new draft, and runs a blind eval on each draft to see whether an LLM judge can spot the synthetic text.
That post puts a number on the claim: Spiral says its generated draft is indistinguishable from a user's real writing 87 percent of the time in its internal eval. The public product site keeps the pitch simpler, promising style matching from personal samples, brand voice references, or favorite writers.
The early reaction in this retweeted user comment is narrower but useful: it points to Spiral's generated style guides as a standout artifact, not just the finished draft. That matters because Spiral is selling voice capture as a reusable asset across future writing, not a one-off prompt trick.
Agent access
Spiral 4.0 also ships as an MCP tool for coding agents. The company's agents.md page says the remote MCP server lives at https://api.writewithspiral.com/mcp/, uses Streamable HTTP, and opens a browser-based OAuth flow on first connection instead of asking users to paste an API key.
Every's launch post says that agent-facing surface now spans three entry points:
- MCP for agent clients
- CLI for terminal workflows
- API for direct product integrations
The same post says Spiral exposes three core actions through CLI and MCP:
write, for draftingpersonalize, for rewriting text in your voicehumanize, for removing common AI tells
That agent-native framing is all over the launch copy. danshipper's main post explicitly names Codex, Claude Code, and OpenClaw as the kind of environments Spiral is targeting, while the official setup doc provides client-specific instructions starting with Claude Code.
Team workspaces
The less flashy ship is team infrastructure. danshipper's launch post says Spiral can now keep an entire 30-person team on brand across landing pages, tweets, podcasts, and marketing emails.
Every's launch post breaks that into shared building blocks. Team workspaces can now hold:
- shared styles
- shared prompts
- shared knowledge
- shared chats
- shared drafts
That turns Spiral from a personal writing assistant into a shared editorial system. The product site had already pitched workspaces and writing styles as core features, but the 4.0 update pushes them closer to a team voice layer that sits on top of both human and agent-written output.
Token pricing
The cleanest concrete change for buyers is price. The retweeted pricing post says Personal drops from $25 to $15, and Every's official launch post says Team drops from $35 to $25 per seat.
The bigger billing change is structural. According to the official launch post, Spiral is moving away from session caps and toward monthly token allotments with pay-as-you-go overages, spend caps, and an option to disable extra usage. Every says the shift was driven by agent usage patterns, where API and MCP workflows can burn through lots of tokens without mapping cleanly to chat-session limits.
That change fits the broader positioning in Every's own ecosystem. This retweeted clip post ties the launch to Dan Shipper's larger line that agents still need human steering, while the launch post says Spiral remains included in Every's $30 bundle alongside Cora, Monologue, Proof, and Sparkle.