Writers' Room
Sometimes one opinion isn’t enough. You’ve got a decision to make about your book — should this subplot stay in, is this chapter pacing right, how should I position this in marketing — and whatever you ask a single agent, you suspect the answer is shaped by that agent’s specific expertise in a way that might be hiding the parts you need to hear. Writers’ Room is the mode that addresses this. Instead of picking one agent to talk to, you broadcast your message to multiple agents simultaneously and each of them responds from their own perspective. Ishvana responds as the orchestrator with big-picture synthesis. Hawken responds as the writing coach with prose-craft input. Lagan responds as the researcher with depth and context. You get three different angles on the same question, side by side, and you use all three to triangulate the answer instead of trusting one.
It’s a chat mode, not a separate panel. You opt into Writers’ Room from the Chat view’s mode selector. Normal chat is one agent at a time. Writers’ Room is multi-agent simultaneous response. You can switch between them at any time and your conversation history preserves both.
How it works
Section titled “How it works”When Writers’ Room is active, sending a message fires three agents in parallel:
- Ishvana — the orchestrator. She responds from her usual perspective: big-picture thinking, cross-referencing across your world, connecting your question to broader project patterns.
- Hawken — the writing coach. He responds with prose-craft input: what the scene needs, what voice choices are working, what’s narratively weak.
- Lagan — the researcher. She responds with research depth: real-world context, relevant precedents, adjacent information you might not have considered.
The three responses arrive simultaneously, each in its own response column (or stacked if screen space is tight). You read all three, compare them, and use whatever combination is useful.
The mode deliberately uses three agents specifically — not all six. Adding GameMaster and WorldKnowledge and Lorekeeper to a Writers’ Room session would dilute the output with perspectives that don’t apply to most questions, and more voices isn’t always better. Three is the right number for multi-perspective without overwhelming the reader.
When to use it
Section titled “When to use it”Writers’ Room is useful for a specific category of question: the kind where you don’t know which angle you need.
Examples:
- “I’m stuck on this chapter — should I cut it or rewrite it?” Ishvana might point out a structural reason to cut (doesn’t serve the arc). Hawken might point out a voice reason to rewrite (the character’s voice is actually working here). Lagan might surface a thematic reason (similar chapters in comp titles are usually kept). Three different angles, one useful decision.
- “How should I approach the final confrontation in book three?” Ishvana looks at your whole project’s trajectory. Hawken focuses on the craft of the scene. Lagan pulls research about how other series have handled similar moments. You get strategic, craft, and reference-level input together.
- “Why isn’t my protagonist working for me right now?” Ishvana thinks about character arc and motivation. Hawken thinks about voice and prose-level characterization. Lagan might search your Legendry for what you actually wrote about the character. All three point at different problems.
- “Is this subplot worth keeping?” Classic multi-angle question. Get structural, craft, and research perspectives at once.
For questions where you already know which agent is the right one — “write me a scene,” “check this for consistency,” “find me research on X” — Writers’ Room is overkill. Use the specific agent. Writers’ Room is specifically for the muddier kind of question where the value is in the disagreement between perspectives.
How the responses differ
Section titled “How the responses differ”Each of the three agents has a different system prompt, different training for how to respond, and different tool access. Their outputs reflect those differences:
- Ishvana’s responses are usually the longest and most synthetic. She connects the question to your broader project, references your Legendry frequently, and tends to offer multiple options with tradeoffs rather than single answers.
- Hawken’s responses are sharper and more craft-focused. He’ll quote prose, evaluate scene-level choices, and give specific writing-craft feedback. He’s less interested in strategic context and more interested in “is this scene working.”
- Lagan’s responses are research-flavored. She’ll pull context from your knowledge base, cite external sources, and evaluate the question against real-world or comparative data. She tends to answer with breadth — “here are several things you might consider” — rather than a single recommendation.
Reading all three at once trains you to pick up on which agent is reliably best for which kind of question. Over time, you’ll notice patterns — “when I’m debating structural changes, Ishvana’s response is the one I usually act on; when I’m debating voice, Hawken’s is the one I trust.” Writers’ Room teaches you your own specialist preferences.
What the responses share
Section titled “What the responses share”Despite their differences, all three agents see the same message, the same conversation history, and the same project context. They’re not given different information — they interpret the same information through different lenses.
This is an important distinction. Writers’ Room isn’t “three agents with three different knowledge bases.” It’s “three agents with three different analytical approaches to the same knowledge base.” If your project’s Legendry is thin, all three agents will struggle equally. If your Legendry is rich, all three will use it differently.
Parallel execution
Section titled “Parallel execution”Under the hood, Writers’ Room fires all three agents concurrently via asyncio.gather() — the same parallel delegation mechanism that Ishvana uses when she fans out to specialists. This means a Writers’ Room response takes roughly as long as the slowest of the three agents, not three times as long as a single response.
The parallel execution is visible in the Etherforce Observability panel — you’ll see three decisions logged in quick succession for a Writers’ Room send, and they’ll be flagged as part of the same parallel request.
Cost considerations
Section titled “Cost considerations”Writers’ Room uses three LLM calls per message instead of one. If you’re running Writers’ Room heavily, your cost tracking numbers will reflect that — expect roughly 3x the per-message cost compared to single-agent chat.
This isn’t a reason to avoid Writers’ Room. The decisions you make in Writers’ Room sessions are usually high-stakes (structural choices, positioning decisions, subplot cuts), and spending extra tokens to get better decision support is a good tradeoff. But you should be aware of the cost, especially if you’re budget-conscious about your LLM spend.
Conversation history
Section titled “Conversation history”A Writers’ Room message and its three responses all persist in the chat history. Switching back to single-agent mode preserves the history — you can scroll up to see the Writers’ Room exchange and any previous single-agent exchanges intermixed.
Future messages in the same conversation can reference earlier Writers’ Room responses. “Ishvana, can you expand on the point you made in your last response about the protagonist’s arc?” works normally — Ishvana sees her own previous turn in the context and can build on it.
What Writers’ Room isn’t
Section titled “What Writers’ Room isn’t”A few honest limits:
- Not a vote. The three agents don’t converge on a single answer. You get three perspectives, and the decision is still yours. If you want agents to debate each other and produce a synthesized conclusion, that’s what the Councils system does — a different feature with a different workflow.
- Not multi-turn debate. Each Writers’ Room send is independent. The agents don’t respond to each other’s answers. They all respond to your message. For actual agent-to-agent debate, use Councils.
- Not more than three agents. GameMaster, WorldKnowledge, and Lorekeeper aren’t in the Writers’ Room by default. Including them would add voices that don’t usually matter for the kind of question Writers’ Room is designed for. If you need a mechanics perspective or a fact-check, ask those agents directly — they’re not excluded from the project, just from this specific mode.
- Not a free feature. See cost considerations above. Writers’ Room is more expensive per message.