bog-agents 0.6.4

Bog Agents - batteries-included agent harness for building AI agents. Supports any major provider (Anthropic, OpenAI, AWS Bedrock, Google, etc.) and local models via Ollama. Built on LangGraph.

Bog Agents

To help you ship LangChain apps to production faster, check out LangSmith. LangSmith is a unified developer platform for building, testing, and monitoring LLM applications.

Quick Install

pip install bog-agents
# or
uv add bog-agents

Getting Started

1. Set your API key (the default model uses Anthropic):

export ANTHROPIC_API_KEY="sk-ant-..."

Or for OpenAI, Google, etc.:

export OPENAI_API_KEY="sk-..."
export GOOGLE_API_KEY="AI..."

2. Create and run an agent:

from bog_agents import create_agent

# Uses Claude Sonnet by default (needs ANTHROPIC_API_KEY)
agent = create_agent()

# Or specify a different provider
agent = create_agent(model="openai:gpt-4o")
agent = create_agent(model="google_genai:gemini-2.0-flash")

# Run the agent
result = agent.invoke(
    {"messages": [{"role": "user", "content": "Hello!"}]},
    config={"configurable": {"thread_id": "my-thread"}},
)

3. Run as an HTTP server (optional):

pip install 'bog-agents[serve]'

from bog_agents import create_agent
from bog_agents.serve import AgentServer

agent = create_agent()
server = AgentServer(agent)
server.run()  # Starts on http://127.0.0.1:8420

What Is This?

Using an LLM to call tools in a loop is the simplest form of an agent. This architecture, however, can yield agents that are "shallow" and fail to plan and act over longer, more complex tasks.

Applications like "Deep Research", "Manus", and "Claude Code" have gotten around this limitation by implementing a combination of four things: a planning tool, sub agents, access to a file system, and a detailed prompt.

bog-agents is a Python package that implements these in a general purpose way so that you can easily create a Bog Agents for your application. For a full overview and quickstart of Bog Agents, the best resource is our docs.

Acknowledgements: This project was primarily inspired by Claude Code, and initially was largely an attempt to see what made Claude Code general purpose, and make it even more so.

Features

Core Agent

• LangGraph-powered — composable agent with middleware architecture
• Multi-model — works with Anthropic, OpenAI, Google, DeepSeek, and any LangChain-compatible LLM
• Sub-agents — spawn isolated sub-agents for complex tasks
• File system — read, write, edit, search files with pluggable backends
• Planning — built-in todo list and plan mode for structured task execution

Middleware (Pluggable)

Middleware	Description
FilesystemMiddleware	File read/write/edit/search + multi-edit and batch read
GitToolsMiddleware	9 git tools: status, diff, log, commit, add, branch, stash, blame, show
RepoMapMiddleware	Structural code map with symbol extraction (Python, JS, TS, Rust, Go, Java)
CheckpointingMiddleware	Git-based snapshots before mutations, with undo/diff
CostTrackerMiddleware	Token usage, cost estimation, budget enforcement, effort levels
PlanModeMiddleware	Read-only mode that blocks mutating tools
AutoQualityMiddleware	Auto-lint/test after edits with project detection
ArchitectMiddleware	Dual-model architect/reviewer with cross-model consultation
ParallelAgentsMiddleware	Concurrent sub-agent task execution
LifecycleHooksMiddleware	15 event types for external tool integration
ContextPackingMiddleware	Smart structured context compression
SummarizationMiddleware	Auto-summarization when context window fills
MemoryMiddleware	Persistent AGENTS.md memory across sessions
SkillsMiddleware	Custom skill/instruction loading
SafeToolsConfig	Per-tool auto-approval rules

Security

• OS-level sandbox — bubblewrap (Linux), seatbelt (macOS), landlock isolation
• Human-in-the-loop — configurable tool approval with interrupt handling

Quick Start

from bog_agents import create_agent

agent = create_agent(
    model="anthropic:claude-sonnet-4-6",
    enable_git_tools=True,
    enable_repo_map=True,
    enable_checkpointing=True,
    enable_cost_tracking=True,
    enable_plan_mode=True,
    auto_lint=True,
    working_dir="/path/to/project",
)

# Run the agent
result = await agent.ainvoke(
    {"messages": [{"role": "user", "content": "Fix the failing tests"}]},
    config={"configurable": {"thread_id": "my-session"}},
)

Resources

• Documentation — Full documentation
• API Reference — Full SDK reference documentation
• Chat LangChain - Chat interactively with the docs

Releases & Versioning

See our Releases and Versioning policies.

Contributing

As an open-source project in a rapidly developing field, we are extremely open to contributions, whether it be in the form of a new feature, improved infrastructure, or better documentation.

For detailed information on how to contribute, see the Contributing Guide.