tinycode CLAUDE.md

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Commands

Note: End users install tinycode via curl -fsSL https://raw.githubusercontent.com/bobbyjohnstx/tinycode/main/install.sh | sh or npx tinycode-ai. The bun install path below is for development only.

# Install dependencies (from repo root, development only)
bun install

# Run in development (TUI mode)
bun dev
bun dev <directory>     # run against a different directory
bun dev .               # run against repo root

# Other dev modes
bun dev serve           # headless API server (port 4096)
bun dev web             # server + open web interface
bun dev acp             # Agent Client Protocol mode (IDE integration, stdio transport)
bun run --cwd packages/app dev        # web UI only (requires server running)
bun run --cwd packages/desktop dev    # Electron desktop app

# Lint
bun run lint            # oxlint across all packages

# Type check (run from a package directory, not root)
bun typecheck           # from e.g. packages/tinycode

# Tests (run from a package directory, never from root)
bun test                # from e.g. packages/tinycode
bun test --timeout 30000 path/to/file.test.ts  # single test file

# Build standalone binary
./packages/tinycode/script/build.ts --single

# Regenerate SDK after API changes (server.ts changes → run this)
./script/generate.ts

# Export session to JSON or HTML
tinycode export --format json <session-id>       # JSON format (default)
tinycode export --format html <session-id>       # Self-contained HTML file

Tests cannot run from repo root (do-not-run-tests-from-root guard). Always cd into a package first.

Testing

Mock LLM Provider

For deterministic testing of session logic, use MockLanguageModel from test/fake/mock-language-model.ts. It accepts a sequence of scenarios that define responses for successive LLM calls:

import { MockLanguageModel, type MockScenario } from "@/test/fake/mock-language-model"

const scenarios: MockScenario[] = [
  { type: "text", content: "First response" },
  { type: "text", content: "Second response" },
  {
    type: "tool-call",
    calls: [{ id: "call-1", name: "tool-name", args: { key: "value" } }],
  },
  { type: "error", error: new Error("Simulated failure") },
]

const model = new MockLanguageModel(scenarios)