bog-agents-cli 0.6.5

A coding agent in your terminal. 50+ commands, any LLM provider, persistent memory, git workflow, code review, plan mode, remote sandboxes, and CI/CD automation. One install, no code required.

Bog Agents CLI

A coding agent in your terminal. Point it at the problem, step back, let it work.

No scaffolding, no boilerplate, no configuration ceremony. One install and you've got a full-blooded AI agent — file access, shell commands, git workflow, code review, planning, sub-agents, the whole outfit. Runs on any LLM that supports tool calling: Anthropic, OpenAI, AWS Bedrock, Google, Ollama, and a dozen others.

Built on the Bog Agents SDK and LangGraph. MIT licensed.

---

Install

pip install bog-agents-cli

# Pick your provider (OpenAI included by default)
pip install 'bog-agents-cli[anthropic]'
pip install 'bog-agents-cli[bedrock]'        # AWS Bedrock
pip install 'bog-agents-cli[ollama]'         # Local, no API key
pip install 'bog-agents-cli[all-providers]'  # Everything

Or with uv:

uv tool install 'bog-agents-cli[anthropic]'

First Run

bog-agents

If you've got an API key in your environment or AWS credentials in ~/.aws/, it picks them up automatically. No key? The setup wizard walks you through it — 30 seconds and you're riding.

# Or specify your model explicitly
bog-agents -M claude-sonnet-4-6
bog-agents -M gpt-4o
bog-agents -M ollama:llama3            # No API key needed
bog-agents -M bedrock_converse:anthropic.claude-sonnet-4-6  # AWS credentials

Check your setup any time:

bog-agents --doctor

---

Features

Interactive TUI

A rich terminal interface with streaming responses, syntax highlighting, inline diffs, and tool-call approval. Everything happens in the terminal — no browser, no Electron, no nonsense.

Slash Commands

Type / in the interactive session and the autocomplete shows you everything. The list below focuses on the commands that are shipped, wired, and ready to use today:

Command	What It Does
/model	Switch LLM mid-session — Anthropic, OpenAI, Ollama, anything
/plan	Read-only plan mode. Agent sees the lay of the land without touching a thing
/effort	Set reasoning depth: low (fast), medium, high, max (thorough)
/review	Code review on staged changes, a commit, or specific files
/diff	Show pending file changes as unified diffs
/compact	Compress conversation context (aggressive, moderate, or custom rules)
/cost	Real-time token usage, cost estimate, and budget enforcement
/context	Show context window usage with breakdown
/remember	Persist insights to agent memory (survives across sessions)
/agent	Spawn and manage parallel agent threads
/background	Queue local background work and monitor status
/dashboard	Show a live multi-agent snapshot
/worktree	Isolated git worktrees for parallel work streams
/doctor	Health check — Python, packages, API keys, tools, sandbox support
/threads	Browse and resume previous conversations
/recommend	AI-powered code review with persona-based analysis
/onboard	Interactive codebase tour for getting up to speed
/docs	Open documentation and project guides
/changelog	Open the project changelog
/mcp	Show active MCP servers and available tools
/extensions	Manage installed extensions
/plugin	Unified plugin and extension install/list/info/enable/disable flow
/keybindings	Show current keybindings or the config path
/remote	Submit a task for remote execution, including SSH sandbox providers
/profile	Switch configuration presets
/session	Show and persist label, tags, project, summary, and exports
/resume	Resume the latest thread, a specific thread, or threads by tag/project
/clear	Start a fresh thread
/quit	Hang up your hat

Non-Interactive Mode

This is where automation lives. One command, one task, exit code tells the story.

# Basic task — no shell access by default
bog-agents -n 'Summarize the README'

# Grant shell access (safe defaults)
bog-agents -n 'Run the test suite' --shell-allow-list recommended

# Specific commands only
bog-agents -n 'Search logs for errors' --shell-allow-list cat,grep,find

# Full shell access
bog-agents -n 'Fix the failing tests and commit' --shell-allow-list all

# Clean output for piping
bog-agents -p 'Explain this code' < my_file.py

# Pipe to another command
bog-agents -p 'Write a code review' < pr_diff.patch | tee review.md

# Machine-readable JSON
bog-agents -n 'List all TODO comments' --json

# No streaming (buffer full response)
bog-agents -n 'Refactor the auth module' --no-stream

# Fix an issue and open a PR in one shot
bog-agents -n 'Fix issue #42' --pr --shell-allow-list all

# Create a draft PR against a specific branch
bog-agents -n 'Add dark mode' --pr --pr-base develop --pr-draft --shell-allow-list all

Exit codes: 0 success, 1 error, 130 interrupted.

Shell access in non-interactive mode is off by default — you grant it explicitly:

• --shell-allow-list recommended — curated safe commands (ls, cat, grep, find, wc, etc.)
• --shell-allow-list ls,cat,grep — your own allow-list
• --shell-allow-list all — unrestricted shell (use in trusted environments)

Conversation Resume

Pick up where you left off. Every conversation is a thread with full history.

bog-agents -r              # Resume most recent thread
bog-agents -r abc123       # Resume a specific thread
bog-agents threads list    # See all threads
bog-agents threads delete abc123  # Clean up

Persistent Memory

The agent remembers things across sessions. Use /remember to persist insights, and use /session to attach labels, tags, project names, summaries, and exports to a thread. Memory is stored per-agent in ~/.bog-agents/.

Custom Skills

Extend the agent with reusable skills and packaged extensions. Skills are instruction bundles with a SKILL.md manifest, and extensions can contribute both skills and slash commands.

bog-agents skills list           # See installed skills
bog-agents skills create         # Scaffold a new skill
bog-agents skills info my-skill  # Show skill details
bog-agents skills delete my-skill

In the interactive TUI:

/plugin
/plugin install <path-or-url>
/plugin info <name>
/plugin enable <name>
/plugin disable <name>

Named Agents

Run multiple agents with separate memory, prompts, and thread history.

bog-agents -a researcher    # Use the "researcher" agent
bog-agents -a reviewer      # Use the "reviewer" agent
bog-agents list             # See all agents
bog-agents reset --agent researcher  # Reset an agent's prompt

Remote Sandboxes

Bog Agents supports both built-in runtime sandboxes and tracked remote tasks. Use the runtime sandbox flags when you want a single interactive or non-interactive session isolated from your local machine, and use /remote when you want queueing, status tracking, or an SSH-backed sandbox workflow.

bog-agents --sandbox modal           # Modal serverless sandbox
bog-agents --sandbox daytona         # Daytona cloud sandbox
bog-agents --sandbox runloop         # Runloop sandbox
bog-agents --sandbox-id existing-id  # Reuse an existing sandbox

Inside the TUI:

/remote config
/remote submit --label scout --branch-prefix fix "investigate the failing tests"
/remote status <id>
/remote stop <id>

MCP (Model Context Protocol)

Load external tools via MCP servers. Auto-discovers .mcp.json in your project, or specify a config file.

bog-agents --mcp-config ./my-mcp-servers.json
bog-agents --no-mcp                  # Disable MCP entirely
bog-agents --trust-project-mcp       # Skip the approval prompt

HTTP API Server

Serve the agent as an HTTP API for integration with other tools.

bog-agents --serve                           # localhost:8420
bog-agents --serve --serve-host 0.0.0.0 --serve-port 9000

ACP Server

Run as an Agent Client Protocol server (for Zed editor integration).

bog-agents --acp

---

Model Configuration

Auto-Detection

The CLI checks for credentials in this order:

1. [models].default in ~/.bog-agents/config.toml
2. [models].recent (last /model switch)
3. ANTHROPIC_API_KEY env var
4. OPENAI_API_KEY env var
5. AWS Bedrock (~/.aws/credentials, AWS_ACCESS_KEY_ID, AWS_PROFILE)
6. GOOGLE_API_KEY env var
7. GOOGLE_CLOUD_PROJECT (Vertex AI)
8. NVIDIA_API_KEY env var
9. Ollama (checks if ollama binary exists)
10. Interactive setup wizard (if nothing found)

Setting a Default

bog-agents --default-model anthropic:claude-sonnet-4-6
bog-agents --default-model                       # Show current default
bog-agents --clear-default-model                 # Remove default

Configuration File

Advanced configuration lives in ~/.bog-agents/config.toml:

[models]
default = "anthropic:claude-sonnet-4-6"

[providers.anthropic]
temperature = 0.7
max_tokens = 8192

[providers.openai]
api_base = "https://my-proxy.example.com/v1"

Runtime Model Parameters

bog-agents -M gpt-4o --model-params '{"temperature": 0.2, "max_tokens": 4096}'
bog-agents -M claude-sonnet-4-6 --profile-override '{"max_input_tokens": 100000}'

---

Supported Providers

Any LangChain-compatible chat model works. Use provider:model format.

Provider	Install Extra	Example
Anthropic	anthropic	anthropic:claude-sonnet-4-6
OpenAI	(included)	openai:gpt-4o
AWS Bedrock	bedrock	bedrock_converse:anthropic.claude-sonnet-4-6
Google AI	google-genai	google_genai:gemini-2.5-pro
Vertex AI	vertexai	google_vertexai:gemini-2.5-pro
Ollama	ollama	ollama:llama3
Groq	groq	groq:llama-3.3-70b
DeepSeek	deepseek	deepseek:deepseek-chat
Fireworks	fireworks	fireworks:llama-v3p3-70b
Mistral	mistralai	mistralai:mistral-large
NVIDIA	nvidia	nvidia:nemotron-70b
OpenRouter	openrouter	openrouter:meta-llama/llama-3
Perplexity	perplexity	perplexity:sonar-pro
xAI	xai	xai:grok-2
Cohere	cohere	cohere:command-r-plus
Together	(via litellm)	litellm:together/llama-3-70b
HuggingFace	huggingface	huggingface:meta-llama/Llama-3
Azure OpenAI	(via openai)	azure_openai:gpt-4o

---

CI/CD & Scripting Recipes

# Code review in CI
git diff main...HEAD | bog-agents -p 'Review this diff for bugs and style issues'

# Generate commit messages
bog-agents -p 'Write a conventional commit message for the staged changes' \
  --shell-allow-list git

# Automated refactoring
bog-agents -n 'Rename getUserData to fetch_user_data across the codebase' \
  --shell-allow-list recommended

# Documentation generation
bog-agents -n 'Generate docstrings for all public functions in src/' \
  --shell-allow-list recommended

# Security audit
bog-agents -n 'Audit this repo for security vulnerabilities' \
  --shell-allow-list recommended --json

# Fix and PR in one shot (great for issue bots)
bog-agents -n 'Fix issue #123' --pr --shell-allow-list all

---

Environment Variables

Variable	Purpose
ANTHROPIC_API_KEY	Anthropic API key
OPENAI_API_KEY	OpenAI API key
AWS_ACCESS_KEY_ID / AWS_PROFILE	AWS Bedrock credentials
GOOGLE_API_KEY	Google AI API key
GOOGLE_CLOUD_PROJECT	Vertex AI project
NVIDIA_API_KEY	NVIDIA API key
TAVILY_API_KEY	Tavily web search
BOG_AGENTS_SHELL_ALLOW_LIST	Default shell allow-list
BOG_AGENTS_LANGSMITH_PROJECT	LangSmith tracing project

Keys can also be set in .env (project-level) or ~/.bog-agents/.env (user-level).

---

Full CLI Reference

bog-agents [OPTIONS] [COMMAND]

Commands:
  list                          List available agents
  reset                         Reset an agent's prompt
  skills                        Manage skills (list/create/info/delete)
  threads                       Manage threads (list/delete)

Core:
  -M, --model MODEL             Model to use
  -a, --agent NAME              Agent name (default: agent)
  -r, --resume [ID]             Resume a thread
  -m, --message TEXT            Auto-submit prompt on start
  --auto-approve                Auto-approve tool calls
  --doctor                      Run diagnostics
  -v, --version                 Show versions
  -h, --help                    Show help

Non-Interactive:
  -n, --non-interactive MSG     Run task and exit
  -p, --print TEXT              Clean output mode (-n + -q)
  -q, --quiet                   Suppress UI chrome
  --no-stream                   Buffer response
  --json                        JSON output
  --shell-allow-list CMDS       Shell access control
  --pr                          Create PR from output
  --pr-base BRANCH              PR base branch
  --pr-draft                    Draft PR

Model:
  --model-params JSON           Extra model kwargs
  --profile-override JSON       Override profile fields
  --default-model [MODEL]       Set/show default model
  --clear-default-model         Clear default

Sandbox:
  --sandbox TYPE                Sandbox provider
  --sandbox-id ID               Reuse existing sandbox
  --sandbox-setup PATH          Setup script

MCP:
  --mcp-config PATH             MCP config file
  --no-mcp                      Disable MCP
  --trust-project-mcp           Trust project MCP

Server:
  --serve                       HTTP API mode
  --serve-host HOST             API host
  --serve-port PORT             API port
  --acp                         ACP server mode

---

Requirements

• Python 3.11+
• At least one LLM provider (API key or local model)

---

Contributing

We're open to contributions. See CONTRIBUTING.md.

License

MIT

---

The trail's marked. Saddle up.