# hunch

> Decision engineering: hunch turns a question into a decision your software can act on, and tells you how often it's right.

- [Introduction](https://fuguai.mintlify.app/index.md): hunch turns a question into a decision your software can act on, and tells you how often it's right.
- [Quickstart](https://fuguai.mintlify.app/quickstart.md): Build a guard for a coding agent's shell commands, measure it, change it and review it, on 38 commands real agents ran. About five minutes and a fifth of a cent.
- [Concepts](https://fuguai.mintlify.app/concepts.md): The ideas hunch is built from, each resting on the one before: rows, questions, specs, the store, gold, act and review.
- [Guard your coding agent](https://fuguai.mintlify.app/guides/guard-your-agent.md): Before Claude Code or Codex runs a shell command, ask whether a person should look first, and keep every decision so you can measure the guard on your own work.
- [Write a spec](https://fuguai.mintlify.app/guides/write-a-spec.md): Turn a judgment you make today into a spec: the rows, what the model sees, the question, and how to test it.
- [Review](https://fuguai.mintlify.app/guides/review.md): How hunch chooses which rows to show you, how to judge them, and where your verdicts go.
- [Test in CI](https://fuguai.mintlify.app/guides/test-in-ci.md): Set thresholds in the spec, and fail the build when a change makes a judgment worse.
- [Share what your judgments decide](https://fuguai.mintlify.app/guides/share.md): Turn a project into one page anyone can read and search: what each judgment decides, how well it was measured, and what depends on it.
- [Change a spec safely](https://fuguai.mintlify.app/guides/change-a-spec.md): See which rows an edit flips before you ship it, try a candidate on live traffic, and decide what happens to old answers.
- [Use a spec in your app](https://fuguai.mintlify.app/guides/use-in-your-app.md): Call the spec you tested from application code, one row at a time, with the same answers and cache as batch runs.
- [Cookbooks](https://fuguai.mintlify.app/cookbooks/index.md): Four real problems worked through with hunch, each with its spec, its numbers, and what did not work.
- [Grade Claude Code sessions](https://fuguai.mintlify.app/cookbooks/claude-code-sessions.md): Read coding-agent sessions as rows, judge how each turn went, and check the judge against reviews.
- [Guard a coding agent's commands](https://fuguai.mintlify.app/cookbooks/command-guard.md): Decide which shell commands a person should see before they run, then measure the guard: how often it stops a harmless command, and how often it lets a harmful one through.
- [Route banking intents](https://fuguai.mintlify.app/cookbooks/banking-intents.md): Sort bank customers' messages into 77 intents, and find out that the answer key is wrong more often than the model.
- [Predict whether a patch passes](https://fuguai.mintlify.app/cookbooks/swe-patches.md): Judge a coding agent's patch without running its tests, and learn that a confident no is worth more than a confident yes.
- [100,000 product reviews](https://fuguai.mintlify.app/cookbooks/amazon-reviews.md): What a 100,000-row run costs, how long it takes, and what sets its speed.
- [Spec reference](https://fuguai.mintlify.app/reference/spec.md): Every key a hunch spec (YAML) accepts, what it does, and one example each.
- [CLI](https://fuguai.mintlify.app/reference/cli.md): Every hunch command and flag, what each line of output means, and the exit codes.
- [Python API](https://fuguai.mintlify.app/reference/python.md): Every function the hunch package exports, with its exact signature and what it returns.
- [Environment variables](https://fuguai.mintlify.app/reference/environment.md): Every environment variable hunch and hunch-server read, with its default and effect.
- [Engines](https://fuguai.mintlify.app/reference/engines.md): The engines a spec's model can name, how an LLM is read through its token probabilities, and how to compare or escalate between them.
- [Projects](https://fuguai.mintlify.app/reference/projects.md): A folder of specs where one judgment reads another's answers: ref, where, chain, union, and how compile, run, test and review treat the graph.
- [Store](https://fuguai.mintlify.app/reference/store.md): The SQLite file that holds every answer, every judgment's table and the history of runs.
- [Server](https://fuguai.mintlify.app/reference/server.md): Serve judgments over HTTP, review in the browser, and watch runs and drift, from the same specs and store as the CLI.
- [Server API](https://fuguai.mintlify.app/reference/server-api.md): Every endpoint of hunch-server: parameters, responses, authentication and errors.
- [How the numbers work](https://fuguai.mintlify.app/reference/how-the-numbers-work.md): Where test's accuracy estimate, interval, calibration, dial and AUROC come from, and how diff decides a change is real.