How to Use Fugu Ultra in opencode
Fugu Ultra is built for difficult coding, research, and reasoning tasks that need several steps. Unlike Fugu, it coordinates multiple expert agents within a request to prioritize answer quality.
Prompts to try with Fugu Ultra in opencode
Copy an example into opencode, or ask in your own words.
Use Fugu Ultra to review the current diff and its callers. Report reproducible bugs with file references, triggering inputs, and a focused regression test.
Use Fugu Ultra to investigate the failing tests in this project. Compare plausible root causes, reproduce the failure, and make the smallest supported fix.
Use Fugu Ultra to turn the research method in this project's documentation into an implementation plan. Identify missing experimental details before editing code.
Use Fugu Ultra to review this project's authentication flow. Trace authorization checks across routes and services, then propose tests for each suspected gap.
Use Fugu Ultra to inspect the experiment logs and training code. Propose one controlled change, run the relevant evaluation, and compare the result with the baseline.
How to set up Fugu Ultra in opencode
Pick the way that suits you. Each takes about two minutes.
Use Spiral's MCP server
- Add the server
Leave out --global to add Spiral to the current project only. No header and no client ID or secret: opencode registers itself with Spiral.
opencode mcp add spiral --global --url https://api.spiral.ai/mcp
- Sign in
Your browser opens. Sign in to Spiral, pick the organization and click Authorize, then return to the terminal. No API key. Inside opencode, /mcps starts the same sign-in.
opencode mcp auth spiral
- Check it
Spiral shows as connected.
opencode mcp list
Use the Spiral skill
- Send opencode this message
Your agent reads Spiral's skill and sets itself up.
set up https://spiral.ai/SKILL.md
- Sign in or paste a key
Approve Spiral in the browser, or give the agent a key from the console.
Use the Spiral CLI
- Install and sign in
Installs the Spiral CLI, signs you in and adds its skill to your coding agents.
npx -y spiral-cli@latest init
Call the Spiral API
- Create an API key
In the console, under API keys.
- Run Fugu Ultra
One request; the result comes back by polling or a webhook.
curl https://api.spiral.ai/v1/runs \ -H "Authorization: Bearer $SPIRAL_API_KEY" \ -H "Content-Type: application/json" \ -H "Idempotency-Key: $(uuidgen)" \ -d '{ "provider": "openrouter", "model": "sakana/fugu-ultra", "input": { "messages": [ { "role": "system" } ] } }'
Then paste one of the prompts above: opencode finds Fugu Ultra on Spiral, shows the price and runs it.
Fugu Ultra providers and prices
Spiral runs Fugu Ultra through OpenRouter. Open a card for today's price, with every fee included.
Up to $0.159972 per run at the default inputs. $6.50 per 1M input tokens, $39.00 per 1M output tokens.
Settings you can ask for
Ask opencode for any of these in your prompt.
| Setting | Allowed values | Default | Description |
|---|---|---|---|
| Reasoning effort | high, xhigh, or max | xhigh | How much the model reasons. |
Every API field of Sakana: Fugu Ultra is on its tool page: open its card above.
What can you do with Fugu Ultra in opencode?
Fugu Ultra can help investigate bugs, review changes, and turn research methods into working code. Give the agent a concrete completion check so it can connect implementation work with test results.
Review a difficult code change
Ask for defects that depend on interactions across functions, with a reproduction case and suggested fix for each finding.
Investigate a persistent bug
Provide the symptoms, relevant code, and test results. Ask for competing explanations and tests that distinguish them.
Plan a paper reproduction
Turn a research method into implementation steps, experiment controls, evaluation criteria, and a list of missing details.
Compare research findings
Compare supplied papers by method, evidence, assumptions, and disagreements. Require a source reference for each conclusion.
Map patent claims
Compare supplied patent claims with technical literature and identify overlapping concepts for further investigation.
Review application security
Analyze supplied authentication logic or security findings. Request evidence, affected paths, and a verification plan.
Design the next research experiment
Use results from previous experiments to propose a testable hypothesis, controlled change, and measurable success criterion.
More on Fugu Ultra and opencode
Overview
Fugu Ultra in other agents
More opencode guides
Questions about Fugu Ultra in opencode
Can opencode use Fugu Ultra?
Yes. Add Spiral to opencode as an MCP server, and opencode can find and run Fugu Ultra like any built-in tool.
How much does Fugu Ultra cost in opencode?
Up to $0.159972 per run at the default inputs. $6.50 per 1M input tokens, $39.00 per 1M output tokens. You pay from your Spiral balance as you go, with no subscription, and a failed run is not charged. Your opencode plan is separate.
Do I need an OpenRouter account to use Fugu Ultra in opencode?
No. Spiral runs Fugu Ultra on its own OpenRouter account. You sign in to Spiral once and never paste a key.
How long does it take to set up Fugu Ultra in opencode?
About two minutes: add the server, then sign in, then check it.
Does opencode wait while Fugu Ultra runs?
Fugu Ultra runs in the background. opencode starts the run, checks on it and gives you the text, json when it's ready.
What is Fugu Ultra best suited for?
Difficult coding, research, and reasoning tasks that require several steps. Sakana reports early use in paper reproduction, cybersecurity analysis, literature reviews, and patent investigations.
How does Fugu Ultra differ from Fugu?
Fugu Ultra can compose a workflow involving multiple expert agents within one request. Its design prioritizes answer quality on difficult problems.
How much context can Fugu Ultra handle?
Fugu Ultra supports a 1,000,000-token context window and up to 128,000 completion tokens. Treat these as separate limits when planning a large review or report.
Does maximum reasoning always add more reasoning than extra high?
Sakana documents those settings as equivalent for the unversioned Fugu Ultra model. Its documentation identifies v1.1 as a release with a distinct maximum level, so the behavior depends on the release.
Does a short output limit restrict all of Fugu Ultra's internal work?
Sakana says the output limit applies to the final response; the orchestrator still uses its own maximum token limit. Use the limit to control answer length.
Start using Fugu Ultra in opencode
One Spiral account for every tool your agent needs. No subscription; you pay per run.
Paste this into opencode:
set up https://spiral.ai/SKILL.md


