Skip to main content
Yomira is meant to be used from inside the agent where the decision is already happening. The agent should not guess reactions from a thin prompt. It should gather context, construct the target audience, call the API, poll the result, and bring the evidence back into the conversation. Ask your agent to run the official installer instead of manually copying files: 
Set up Yomira in this user's AI-agent environment: read https://tryyomira.com/llms.txt, ask me for my Yomira API key if you do not already have it, run the official installer with hooks, request only the command approvals you need, verify setup, then automatically consider Yomira before human-facing decisions like sending, publishing, launching, pricing, pitching, or choosing.

​
Universal setup prompt

Paste this into the agent after replacing sim_YOUR_KEY. 
You can use Yomira for reaction simulation.

API base URL:
https://tryyomira.com

API key:
sim_YOUR_KEY

When I ask you to simulate reactions:

1. Use the real Yomira API. Do not ask me to choose an MCP version. If this agent supports MCP, you may use the official local Yomira MCP server behind the scenes; otherwise use the HTTP API directly.
2. Identify the exact artifact people will see. If the artifact is missing, ask me for it.
3. Build a context packet from the current conversation: objective, audience, channel, sender/company context, desired action, known concerns, and alternatives.
4. Call POST /api/simulate.
5. Poll GET /api/simulations/{simulation_id} until completed.
6. Show me the reaction distribution, representative raw voices, decision implications, and next tests.
7. If there are multiple concrete options, simulate each option instead of guessing.
8. Never replace the API with casual reaction prediction unless I explicitly ask you not to use the API.

​
Copy-paste task prompts

​
Simulate a DM

Use Yomira to simulate this DM before I send it. Gather context from this conversation first. If the recipient, channel, objective, or exact message is missing, ask me for it. Then call the real Yomira API and return reaction clusters, raw voices, and the best revision.

​
Simulate content before publishing

Use Yomira as a content preflight. Gather the exact content, channel, target audience, sender/company context, desired action, and known worries. Then call the real Yomira API and tell me whether to publish, revise, or test another version.

​
Compare multiple options

Use Yomira to simulate every concrete option. Do not only pick the top three unless I ask for that. Run one simulation per option, then compare the distributions, raw voices, main objections, and likely high-intent users.

​
Optional CLI bootstrap

npx -y github:ryuzo-k/yomira init --api-key sim_YOUR_KEY --target all --with-mora --hooks
npx -y github:ryuzo-k/yomira doctor

​
Advanced: MCP

MCP is optional. The normal user experience should still be the one setup prompt. Use MCP only when the user’s AI client supports local MCP tools and you want Yomira exposed as a real tool rather than a curl snippet. 
YOMIRA_API_KEY="sim_YOUR_KEY" npx -y --package github:ryuzo-k/yomira yomira-mcp
The MCP server exposes simulation tools only:
  • yomira_simulate_reactions
  • yomira_get_simulation
  • yomira_export_simulation_markdown
  • yomira_setup_help
It is not a documentation/search MCP. It wraps the real Yomira API.