* feat: migrate Bedrock provider to AI SDK Replace the raw AWS SDK (@aws-sdk/client-bedrock-runtime) Bedrock handler with the Vercel AI SDK (@ai-sdk/amazon-bedrock). Reduces provider from 1,633 lines to 575 lines (65% reduction). Key changes: - Use streamText()/generateText() instead of ConverseStreamCommand/ConverseCommand - Use createAmazonBedrock() with native auth (access key, secret, session, profile via credentialProvider, API key, VPC endpoint as baseURL) - Reasoning config via providerOptions.bedrock.reasoningConfig - Anthropic beta headers via providerOptions.bedrock.anthropicBeta - Thinking signature captured from providerMetadata.bedrock.signature on reasoning-delta stream events - Thinking signature round-tripped via providerOptions.bedrock.signature on reasoning parts in convertToAiSdkMessages() - Redacted thinking captured from providerMetadata.bedrock.redactedData - isAiSdkProvider() returns true for reasoning block preservation - Keep: getModel, ARN parsing, cross-region inference, cost calculation, service tier pricing, 1M context beta Tests: 83 tests skipped (mock old AWS SDK internals, need rewrite for AI SDK mocking). 106 tests pass. 0 tests fail. * fix: address review feedback for Bedrock AI SDK migration - Wire usePromptCache into AI SDK via providerOptions.bedrock.cachePoint on system prompt and last two user messages - Remove debug logger.info that fires on every stream event with providerMetadata - Tighten isThrottlingError to match 'rate limit' instead of broad 'rate'/'limit' substrings that false-positive on context length errors - Use shared handleAiSdkError utility for consistent error handling with status code preservation for retry logic * fix: bedrock AI SDK migration - fix usage metrics, rewrite tests, remove dead code - Fix reasoningTokens always 0 (usage.details?.reasoningTokens → usage.reasoningTokens) - Fix cacheReadInputTokens always 0 (read from usage.inputTokenDetails instead of providerMetadata) - Fix invokedModelId not extracted for prompt router cost calculation - Rewrite all 6 skipped bedrock test suites for AI SDK mocking pattern (140 tests pass) - Remove dead code: bedrock-converse-format.ts, cache-strategy/ (6 files, ~2700 lines) * chore: remove dead @anthropic-ai/bedrock-sdk dep and stale AWS SDK mocks * chore: update pnpm-lock.yaml after removing @anthropic-ai/bedrock-sdk * fix: compute cache point indices from original Anthropic messages before AI SDK conversion The previous approach naively targeted the last 2 user messages in the post-conversion AI SDK array, but convertToAiSdkMessages() splits user messages containing tool_results into separate tool + user messages, causing cache points to land on the wrong messages (tiny text fragments instead of the intended meaty user turns). Now we identify the last 2 user messages in the original Anthropic message array (matching the Anthropic provider's caching strategy) and build a parallel-walk mapping to apply cachePoint to the correct corresponding AI SDK message. * perf: optimize prompt caching with 3-point message strategy + anchor for 20-block window Previous approach only cached the last 2 user messages (using 2 of 4 available cache checkpoints for messages). This left significant cache savings on the table for longer conversations. New strategy uses up to 3 message cache points (+ 1 system = 4 total): - Last user message: write to cache for next request - Second-to-last user message: read from cache for current request - Anchor message at ~1/3 position: ensures the 20-block lookback window from the second-to-last breakpoint hits a stable cache entry, covering all assistant/tool messages in the middle of the conversation Also extracted the parallel-walk mapping logic into a reusable applyCachePointsToAiSdkMessages() helper method. Industry benchmarks show 70-95% token cache rates are achievable; this change should significantly improve our 39% baseline for longer multi-turn conversations. * chore: remove stale bedrock-sdk external, fix arnInfo property name, remove unused exports --------- Co-authored-by: daniel-lxs <ricciodaniel98@gmail.com> |
||
|---|---|---|
| .. | ||
| docs | ||
| scripts | ||
| src | ||
| CHANGELOG.md | ||
| eslint.config.mjs | ||
| install.sh | ||
| package.json | ||
| README.md | ||
| tsconfig.json | ||
| tsup.config.ts | ||
| vitest.config.ts | ||
@roo-code/cli
Command Line Interface for Roo Code - Run the Roo Code agent from the terminal without VSCode.
Overview
This CLI uses the @roo-code/vscode-shim package to provide a VSCode API compatibility layer, allowing the main Roo Code extension to run in a Node.js environment.
Installation
Quick Install (Recommended)
Install the Roo Code CLI with a single command:
curl -fsSL https://raw.githubusercontent.com/RooCodeInc/Roo-Code/main/apps/cli/install.sh | sh
Requirements:
- Node.js 20 or higher
- macOS Apple Silicon (M1/M2/M3/M4) or Linux x64
Custom installation directory:
ROO_INSTALL_DIR=/opt/roo-code ROO_BIN_DIR=/usr/local/bin curl -fsSL ... | sh
Install a specific version:
ROO_VERSION=0.1.0 curl -fsSL https://raw.githubusercontent.com/RooCodeInc/Roo-Code/main/apps/cli/install.sh | sh
Updating
Re-run the install script to update to the latest version:
curl -fsSL https://raw.githubusercontent.com/RooCodeInc/Roo-Code/main/apps/cli/install.sh | sh
Uninstalling
rm -rf ~/.roo/cli ~/.local/bin/roo
Development Installation
For contributing or development:
# From the monorepo root.
pnpm install
# Build the main extension first.
pnpm --filter roo-cline bundle
# Build the cli.
pnpm --filter @roo-code/cli build
Usage
Interactive Mode (Default)
By default, the CLI prompts for approval before executing actions:
export OPENROUTER_API_KEY=sk-or-v1-...
roo "What is this project?" -w ~/Documents/my-project
You can also run without a prompt and enter it interactively in TUI mode:
roo -w ~/Documents/my-project
In interactive mode:
- Tool executions prompt for yes/no approval
- Commands prompt for yes/no approval
- Followup questions show suggestions and wait for user input
- Browser and MCP actions prompt for approval
Non-Interactive Mode (-y)
For automation and scripts, use -y to auto-approve all actions:
roo "Refactor the utils.ts file" -y -w ~/Documents/my-project
In non-interactive mode:
- Tool, command, browser, and MCP actions are auto-approved
- Followup questions show a 60-second timeout, then auto-select the first suggestion
- Typing any key cancels the timeout and allows manual input
Roo Code Cloud Authentication
To use Roo Code Cloud features (like the provider proxy), you need to authenticate:
# Log in to Roo Code Cloud (opens browser)
roo auth login
# Check authentication status
roo auth status
# Log out
roo auth logout
The auth login command:
- Opens your browser to authenticate with Roo Code Cloud
- Receives a secure token via localhost callback
- Stores the token in
~/.config/roo/credentials.json
Tokens are valid for 90 days. The CLI will prompt you to re-authenticate when your token expires.
Authentication Flow:
┌──────┐ ┌─────────┐ ┌───────────────┐
│ CLI │ │ Browser │ │ Roo Code Cloud│
└──┬───┘ └────┬────┘ └───────┬───────┘
│ │ │
│ Open auth URL │ │
│─────────────────>│ │
│ │ │
│ │ Authenticate │
│ │─────────────────────>│
│ │ │
│ │<─────────────────────│
│ │ Token via callback │
│<─────────────────│ │
│ │ │
│ Store token │ │
│ │ │
Options
| Option | Description | Default |
|---|---|---|
[prompt] |
Your prompt (positional argument, optional) | None |
--prompt-file <path> |
Read prompt from a file instead of command line argument | None |
-w, --workspace <path> |
Workspace path to operate in | Current directory |
-p, --print |
Print response and exit (non-interactive mode) | false |
-e, --extension <path> |
Path to the extension bundle directory | Auto-detected |
-d, --debug |
Enable debug output (includes detailed debug information, prompts, paths, etc) | false |
-y, --yes, --dangerously-skip-permissions |
Auto-approve all actions (use with caution) | false |
-k, --api-key <key> |
API key for the LLM provider | From env var |
--provider <provider> |
API provider (roo, anthropic, openai, openrouter, etc.) | openrouter (or roo if authenticated) |
-m, --model <model> |
Model to use | anthropic/claude-opus-4.6 |
--mode <mode> |
Mode to start in (code, architect, ask, debug, etc.) | code |
-r, --reasoning-effort <effort> |
Reasoning effort level (unspecified, disabled, none, minimal, low, medium, high, xhigh) | medium |
--ephemeral |
Run without persisting state (uses temporary storage) | false |
--oneshot |
Exit upon task completion | false |
--output-format <format> |
Output format with --print: text, json, or stream-json |
text |
Auth Commands
| Command | Description |
|---|---|
roo auth login |
Authenticate with Roo Code Cloud |
roo auth logout |
Clear stored authentication token |
roo auth status |
Show current authentication status |
Environment Variables
The CLI will look for API keys in environment variables if not provided via --api-key:
| Provider | Environment Variable |
|---|---|
| roo | ROO_API_KEY |
| anthropic | ANTHROPIC_API_KEY |
| openai-native | OPENAI_API_KEY |
| openrouter | OPENROUTER_API_KEY |
| gemini | GOOGLE_API_KEY |
| vercel-ai-gateway | VERCEL_AI_GATEWAY_API_KEY |
Authentication Environment Variables:
| Variable | Description |
|---|---|
ROO_WEB_APP_URL |
Override the Roo Code Cloud URL (default: https://app.roocode.com) |
Architecture
┌─────────────────┐
│ CLI Entry │
│ (index.ts) │
└────────┬────────┘
│
▼
┌─────────────────┐
│ ExtensionHost │
│ (extension- │
│ host.ts) │
└────────┬────────┘
│
┌────┴────┐
│ │
▼ ▼
┌───────┐ ┌──────────┐
│vscode │ │Extension │
│-shim │ │ Bundle │
└───────┘ └──────────┘
How It Works
-
CLI Entry Point (
index.ts): Parses command line arguments and initializes the ExtensionHost -
ExtensionHost (
extension-host.ts):- Creates a VSCode API mock using
@roo-code/vscode-shim - Intercepts
require('vscode')to return the mock - Loads and activates the extension bundle
- Manages bidirectional message flow
- Creates a VSCode API mock using
-
Message Flow:
- CLI → Extension:
emit("webviewMessage", {...}) - Extension → CLI:
emit("extensionWebviewMessage", {...})
- CLI → Extension:
Development
# Run directly from source (no build required)
pnpm dev --provider roo --api-key $ROO_API_KEY --print "Hello"
# Run tests
pnpm test
# Type checking
pnpm check-types
# Linting
pnpm lint
By default the start script points ROO_CODE_PROVIDER_URL at http://localhost:8080/proxy for local development. To point at the production API instead, override the environment variable:
ROO_CODE_PROVIDER_URL=https://api.roocode.com/proxy pnpm dev --provider roo --api-key $ROO_API_KEY --print "Hello"
Releasing
Official releases are created via the GitHub Actions workflow at .github/workflows/cli-release.yml.
To trigger a release:
- Go to Actions → CLI Release
- Click Run workflow
- Optionally specify a version (defaults to
package.jsonversion) - Click Run workflow
The workflow will:
- Build the CLI on all platforms (macOS Apple Silicon, Linux x64)
- Create platform-specific tarballs with bundled ripgrep
- Verify each tarball
- Create a GitHub release with all tarballs attached
Local Builds
For local development and testing, use the build script:
# Build tarball for your current platform
./apps/cli/scripts/build.sh
# Build and install locally
./apps/cli/scripts/build.sh --install
# Fast build (skip verification)
./apps/cli/scripts/build.sh --skip-verify