Official CLI · Terminal AI Agent · v0.4

BCCLI — Autonomous AI agent
directly in your terminal.

BCCLI is BotConnector's terminal coding agent. It deeply reads codebases, plans multi-file refactoring, executes shell commands with your permission, and connects to both cloud models and local Ollama runtimes in one consistent environment.

TERMINAL INSTALL npm install -g @botconnector/bccli@next

Node.js 22+ Required

Full Git & Diff Safety

Cloud + Local Ollama Routing

MCP Extensibility

TERMINAL WORKFLOW

Experience interactive agentic engineering

BCCLI works alongside you inside your project directory. From quick script generation to large-scale architectural refactors, BCCLI manages context so you can stay in flow.

bash — bccli (workspace-agent)

Ready · Qwen 2.5 Coder 14B

$ cd my-saas-project && bccli

◇ BotConnector CLI v0.4.0-beta.2

Indexed 48 repository files · Node.js v22.14.0 · Git branch: main

> Add rate-limiting middleware to our Express router and run the unit tests.

◇ Analyzing src/server.ts and src/routes/api.ts...

+ import rateLimit from 'express-rate-limit';

+ const limiter = rateLimit({ windowMs: 15 * 60 * 1000, max: 100 });

+ app.use('/api/', limiter);

✓ 14 unit tests passed (0 failed) in 1.42s

CORE CAPABILITIES

Four reasons developers build with BCCLI

Deep Codebase Awareness

BCCLI automatically navigates your repo tree, locates relevant symbols across files, reads configuration files, and understands your project structure before suggesting code.

Safe Permissions & Undo

Never fear broken builds. BCCLI displays visual diffs before writing changes to disk, asks for confirmation on critical shell commands, and supports instant /undo rolls.

Hybrid Cloud & Local Switch

Switch models on the fly. Use local Ollama models (bccli -m local/qwen2.5-coder:14b) for private proprietary work, or cloud flagship models for difficult reasoning.

Model Context Protocol (MCP)

Connect BCCLI to your external databases, internal APIs, GitHub PR trackers, or file systems with Model Context Protocol servers configured in simple JSON.

GET STARTED

Installation & quickstart

Install via NPM

Ensure you have Node.js 22 or newer installed (run node -v to verify).

npm install -g @botconnector/bccli@next

Connect Your Account

Authenticate with your BotConnector Cloud API key, or use your local Ollama runtime without an account.

bccli login          # Save your BotConnector API key

Launch in Any Project

Navigate to your repository and launch an interactive session.

cd ~/my-project
bccli

COMMAND REFERENCE

BCCLI CLI command cheatsheet

Command	Usage & Description	Mode
`bccli`	Start an interactive agentic session in current directory	Interactive
`bccli -p "task"`	Execute a non-interactive task directly (useful for scripts & CI/CD)	Headless / Script
`bccli -c` / `bccli -r`	Resume the previous session or choose from session history	Session continuity
`bccli -m <model>`	Select a specific cloud or local model (e.g., `bc-cloud/glm-5.3-flash`)	Model routing
`bccli models`	List all detected cloud and local models with availability status	Inspection
`bccli login [provider]`	Save an API key for BotConnector Cloud or custom providers	Auth
`bccli mcp list\|add\|remove`	Manage configured Model Context Protocol tool servers	Extensibility
`/doctor`	Run inside a session to inspect runtime health, API connectivity, and permissions	In-session command

BCCLI — Autonomous AI agentdirectly in your terminal.