# TrueCourse

> TrueCourse documentation: AI architecture and code intelligence. Static + LLM code analysis and spec-bound scenario tests that catch business-logic drift.

## Docs

- [Baselines & diff](https://docs.truecourse.dev/analyze/baseline-and-diff.md): Commit a baseline once, then see only what your changes introduce.
- [Excluding files](https://docs.truecourse.dev/analyze/excluding-files.md): What analyze skips automatically, and how to exclude tracked paths with .truecourseignore.
- [Git hooks](https://docs.truecourse.dev/analyze/git-hooks.md): Block commits that introduce new violations at or above a configured severity.
- [Overview](https://docs.truecourse.dev/analyze/overview.md): Static + LLM analysis of your code: architecture, security, bugs, performance, and more.
- [Rules](https://docs.truecourse.dev/analyze/rules.md): Configure rule categories, LLM rules, and individual rules per repository.
- [LLM transport](https://docs.truecourse.dev/configuration/llm-transport.md): How TrueCourse reaches the model: Claude Code, a provider API, or an agent mailbox.
- [Models & environment](https://docs.truecourse.dev/configuration/models.md): Per-stage model selection, concurrency, timeouts, and Claude Code tuning.
- [Storage](https://docs.truecourse.dev/configuration/storage.md): Everything is plain JSON files: what lives where, and what to commit.
- [Telemetry](https://docs.truecourse.dev/configuration/telemetry.md): What anonymous usage data is collected, and how to opt out.
- [Dashboard](https://docs.truecourse.dev/dashboard.md): One web UI for both capabilities: code findings and business-logic drift side by side.
- [Resolving conflicts](https://docs.truecourse.dev/guard/conflicts.md): Review and resolve flagged within-area doc overlaps: pick a side or dismiss.
- [External services](https://docs.truecourse.dev/guard/external-services.md): Hermetic by default: app-owned fakes, scripted stubs, or real sandbox accounts behind a fault-scriptable proxy.
- [Guard generate](https://docs.truecourse.dev/guard/generate.md): Author spec-section-bound scenario tests: classify → generate → birth-validate.
- [Overview](https://docs.truecourse.dev/guard/overview.md): Turn your docs into deterministic scenario tests that catch business-logic drift.
- [The recipe](https://docs.truecourse.dev/guard/recipe.md): scenarios/recipe.json tells guard how to build your repo and what the scenarios drive: a CLI binary, an HTTP service, or both.
- [Guard run](https://docs.truecourse.dev/guard/run.md): Run the committed scenarios deterministically: a CI gate with no LLM in the loop.
- [Seeding](https://docs.truecourse.dev/guard/seeding.md): api.seed mints the rows, fixtures, and authenticated principals your scenarios need, once per run.
- [Guard setup](https://docs.truecourse.dev/guard/setup.md): Prepare the repo for guard: derive and prove the recipe, declare external APIs, draft the data + auth seed.
- [Spec scan](https://docs.truecourse.dev/guard/spec-scan.md): Curate scattered docs into a corpus of areas, doc relations, and overlap flags.
- [Web sources](https://docs.truecourse.dev/guard/web-sources.md): Register llms.txt documentation sites as spec docs, snapshotted into the repo and scanned offline.
- [Introduction](https://docs.truecourse.dev/index.md): TrueCourse catches code defects and business-logic drift: two independent tools over one local, file-based store.
- [Installation](https://docs.truecourse.dev/installation.md): Prerequisites, the npm package, and the first-run LLM choice.
- [Quickstart](https://docs.truecourse.dev/quickstart.md): Install TrueCourse, run your first code analysis, and guard your specs in minutes.
- [Claude Code skills](https://docs.truecourse.dev/reference/claude-code-skills.md): Conversational analysis from within Claude Code.
- [CLI reference](https://docs.truecourse.dev/reference/cli.md): Every truecourse command and flag.