AI-powered coding agent — runs GPT, Claude, Gemini, DeepSeek, and 200+ models in your terminal.
npm install -g @simpletoolsindia/beastcli
beastcli --version # verify installationRequires Node.js 20+.
# Start interactive setup (auto-detects your provider)
beastcli
# Or set up via environment variables
OPENAI_API_KEY=sk-... beastcli -p "write a hello world in python"BeastCLI works with any OpenAI-compatible API. Set one of these:
# Install: https://ollama.ai
ollama pull qwen2.5-coder:7b
beastcli -p "write hello world" \
--provider ollama \
--bareOr with environment variables:
BEASTCLI_PROVIDER=ollama \
OPENAI_BASE_URL=http://localhost:11434/v1 \
OPENAI_MODEL=qwen2.5-coder:7b \
beastcli -p "write hello world" --bareGet a key at openrouter.ai/keys, then:
OPENAI_API_KEY=sk-or-v1-... \
OPENAI_BASE_URL=https://openrouter.ai/api/v1 \
OPENAI_MODEL=google/gemma-3-4b-it \
beastcli -p "write hello world" --bareFree models on OpenRouter: google/gemma-3-4b-it, deepseek/deepseek-chat-v3-0324:free, qwen/qwen-2.5-72b-instruct:free
Get a key at ngc.nvidia.com, then:
OPENAI_API_KEY=nvapi-... \
OPENAI_BASE_URL=https://integrate.api.nvidia.com/v1 \
OPENAI_MODEL=google/gemma-3-4b-it \
beastcli -p "write hello world" --bare# OpenAI
OPENAI_API_KEY=sk-... beastcli -p "write hello world" --bare
# Anthropic
ANTHROPIC_API_KEY=sk-ant-... beastcli -p "write hello world" --bare
# Gemini (OpenAI-compatible endpoint)
GOOGLE_API_KEY=... \
OPENAI_BASE_URL=https://generativelanguage.googleapis.com/v1beta/openai \
OPENAI_MODEL=gemini-2.0-flash \
beastcli -p "write hello world" --bare| Provider | Base URL | API Key Env Var |
|---|---|---|
| Ollama | http://localhost:11434/v1 |
None |
| OpenRouter | https://openrouter.ai/api/v1 |
OPENAI_API_KEY |
| NVIDIA NIM | https://integrate.api.nvidia.com/v1 |
OPENAI_API_KEY |
| OpenAI | https://api.openai.com/v1 |
OPENAI_API_KEY |
| Anthropic | (default) | ANTHROPIC_API_KEY |
| DeepSeek | https://api.deepseek.com/v1 |
OPENAI_API_KEY |
| Groq | https://api.groq.com/openai/v1 |
OPENAI_API_KEY |
| Together AI | https://api.together.xyz/v1 |
OPENAI_API_KEY |
| LM Studio | http://localhost:1234/v1 |
None |
| Custom | Any OpenAI-compatible | OPENAI_API_KEY |
beastcli # full interactive REPL
beastcli --provider ollama # specify provider at startup
beastcli --model llama3.2 # specify modelbeastcli -p "write hello world in python" # single prompt, print output
beastcli -p "explain this: $(cat file.py)" # pipe input
echo "fix this bug" | beastcli -p # stdin input
beastcli -p --model gpt-4o "analyze this code" # specify model# Minimal mode: skips UI, auth, plugins — for scripts and CI
beastcli -p "write tests" --bare
# Chain prompts
beastcli -p "plan the feature" --bare > plan.txt
beastcli -p "implement based on plan" --bareThese run locally without any API key:
| Tool | Example | Description |
|---|---|---|
| Calculator | /calc 2+2 |
Math operations |
| UUID | /uuid |
UUID v1/v4/v7 |
| Hash | /hash sha256 "text" |
MD5, SHA, etc. |
| Base64 | /base64 encode "hi" |
Encode/decode |
| JSON | /json format '{}' |
Format/validate |
| QR Code | /qr "https://..." |
Generate QR |
| Color | /color #FF0000 |
HEX/RGB/HSL |
| URL | /url parse "..." |
Parse/encode URLs |
| Regex | /regex "[a-z]+" "abc" |
Test regex |
| Password | /password 16 |
Secure passwords |
| Text Diff | /diff "a" "b" |
Compare text |
| Text Stats | /stats "text" |
Word count, etc. |
| Lorem | /lorem 5 |
Placeholder text |
| Cron | /cron "0 9 * * *" |
Parse cron |
| Timezone | /tz Tokyo |
Time zones |
| Currency | /currency 100 USD EUR |
Conversion |
| IP Lookup | /ip 8.8.8.8 |
Geolocation |
| Weather | /weather Tokyo |
Current weather |
| News | /news tech |
Tech news |
| Timestamp | /timestamp |
Unix time |
| YouTube | /transcript "url" |
Video transcripts |
| Command | Description |
|---|---|
/help |
Show help |
/clear |
Clear chat |
/config |
Open settings UI |
/provider |
Manage providers |
/doctor |
Run diagnostics |
/commit |
Git commit wizard |
/diff |
Show changes |
/plan |
Structured planning |
/task |
Task management |
/agent |
Spawn sub-agent |
/compact |
Compact context |
/mcp add |
Add MCP server |
/skills |
List/install skills |
| Key | Action |
|---|---|
Ctrl+C |
Cancel operation |
Ctrl+D |
Exit BeastCLI |
Tab |
Autocomplete |
↑/↓ |
Command history |
Ctrl+L |
Clear terminal |
Settings are stored at:
- macOS/Linux:
~/.beastcli/config.json - Windows:
%USERPROFILE%/.beastcli/config.json
To reset to defaults:
rm ~/.beastcli/config.json
beastcli # fresh setup"command not found" after install:
# Find where it was installed
which beastcli
# If using nvm, run:
npm install -g @simpletoolsindia/beastcli
hash -rOllama not connecting:
ollama serve # ensure server is running
curl http://localhost:11434 # should return versionRate limited (NVIDIA/OpenAI):
- Wait 60 seconds and retry
- Use a local model (Ollama) to avoid limits
Context too long:
/compact # compact conversation contextgit clone https://github.com/simpletoolsindia/beastcli
cd beastcli
npm install # or: bun install
bun run build # build dist/cli.mjs
bun test # run test suiteDebug build errors:
bun run build 2>&1 | grep -i errorMIT — see LICENSE
Built by SimpleTools India
GitHub ·
NPM ·
Discord