Bantam: tiny, powerful, DIY AI agent

About

Bantam is a minimalist, dependency-free AI agent specification with two reference implementations: Python (bantam.py, ~300 SLOC) and Go (main.go + term_*.go, module code.luxferre.top/luxferre/bantam). It provides an agentic loop capable of autonomous tool execution, shell interaction, real-time response streaming, Fibonacci backoff network resilience, and subagent delegation using any OpenAI-compatible completions API.

The entire philosophy of Bantam is built upon two principles:

1. The structure must be as simple as possible for anyone to be able to reimplement the agent from a plain algorithm description.
2. The agent only needs to provide two tools: a tool to call shell commands and a tool to call itself. In theory, this should be sufficient to give LLMs the ability to handle tasks of any complexity.

Because of the second principle, Bantam itself was named after Victorinox Bantam Alox, a small and lightweight Swiss army knife with only two tools.

Usage

Prerequisites

• Go 1.21+ (Go implementation, no external dependencies) or Python 3.7+ (reference implementation)
• An OpenAI-compatible API endpoint (or OpenAI API key)

Installation (Go)

go install code.luxferre.top/luxferre/bantam@latest

This installs the bantam binary into $(go env GOPATH)/bin (make sure it is on your PATH). To build from a local checkout instead:

go build ./...        # produces ./bantam
# or run without building:
go run . prompt.txt

The Go port is a single main.go plus two small platform files (term_linux.go, term_darwin.go, term_windows.go, term_other.go) for the built-in raw-terminal line editor — zero external dependencies, same as the Python version.

Running Bantam

Both implementations read the same model.cfg and system.txt from the current working directory.

1. Configure model.cfg with your API settings:

   endpoint=https://api.openai.com/v1
   model=gpt-4o
   temperature=0.7
   api_key=your_api_key_here
   stream=true
   

2. Interactive mode:

   bantam            # Go (or: go run .)
   python3 bantam.py  # Python
   

   In interactive mode, prompts can span multiple lines: press Ctrl+J to insert a real line break (the cursor moves to the next line), then Enter to submit the whole multi-line prompt. The Go port ships its own raw-mode line editor (arrow keys move the cursor, Up/Down browse history, Backspace edits, Ctrl+C/Ctrl+D exit), so this works everywhere without dependencies; the Python port uses readline when available and falls back to single-line prompts otherwise.

   Sessions are saved under ~/.bantam/sessions/ and can be managed with these commands:

   • /save — save the entire conversation to a new session file (auto-id like 20260808-190038) and generate its summary
   • /list — list saved sessions (newest first) with their ids, timestamps, message counts and summaries
   • /load <id> — load a saved session (exact id or unique prefix) and continue from there
   • /compact — summarize the conversation with the LLM and compact the context down to just the system message plus the summary
   • /help — show all supported commands
   • /clear — reset the conversation to just the system prompt
   • /quit — exit

   The current conversation is also auto-saved to ~/.bantam/sessions/autosave.json after every turn, on /clear, /load, /compact, and on exit — so you can always /load autosave to resume where you left off.

3. File input mode:

   python3 bantam.py prompt.txt   # Python
   bantam prompt.txt              # Go
   

Rules of Bantam (The Algorithm)

Using these rules, everyone can build their own copy of Bantam from scratch in little time in any language that supports file access, shell and HTTP(S) calls.

High-Level Overview

1. Initialization: Read system.txt and model.cfg. Prepare an array of messages starting with the system prompt {"role": "system", "content": system_prompt}.
2. Input Processing: Take user prompt (via command-line file parameter or interactive stdin), append {"role": "user", "content": prompt}, and invoke AL(cfg, messages).
3. Agentic Loop (AL):
   • Send messages and tool definitions to the OpenAI-compatible /chat/completions API endpoint with custom User-Agent headers.
   • On network or HTTP failure, retry using Fibonacci backoff delays (1s, 1s, 2s, 3s, 5s).
   • If stream=true, parse SSE data chunks (data: {...}) in real-time to stream reasoning content (reasoning_content) and response text directly to stdout, bracketing the reasoning block with --- reasoning start --- / --- reasoning end --- markers.
   • Reconstruct the assistant message. If tool_calls exist, trace the call ([tool call: name(args)]), validate JSON arguments, execute the requested tool (shell_exec or run_subagent), trace the result ([tool result: name]), append the tool response {"role": "tool", "tool_call_id": id, "content": result}, and repeat the loop.
   • If no tool calls remain or max_al_iterations is reached, return the updated messages list.

---

Main program

1. Read system prompt from system.txt (default if missing).
2. Read model parameters from model.cfg (key=value format).
3. Prepare a new message list with the system prompt (role: "system").
4. Read the first command-line parameter. If non-empty, read user prompt from the specified file, append to messages (role: "user"), run AL(cfg, messages), and exit.
5. Read user prompt from standard input (with readline line editing and history in ~/.bantam_history; Ctrl+J inserts a real newline into the line being edited). If equal to /quit or EOF, exit. If equal to /clear, reset messages to step 3 and return to step 5. If equal to /save, write the whole messages array to ~/.bantam/sessions/<id>.json (with an auto-generated summary) and return to step 5. If equal to /list, print saved sessions and their summaries and return to step 5. If starting with /load, replace messages with the saved session's messages (by exact id or unique prefix) and return to step 5. If equal to /compact, ask the LLM to summarize the conversation, replace messages with [system, summary-user-message], and return to step 5. If equal to /help, print the command list and return to step 5. After every user turn and on exit, auto-save messages to ~/.bantam/sessions/autosave.json.
6. Append user prompt to messages (role: "user"), run AL(cfg, messages), and go to step 5.

Agentic loop (AL(cfg, messages)) function

1. Call OpenAI-compatible Completions API (POST {endpoint}/chat/completions) using parameters from cfg (model, temperature, optional api_key bearer header).
   • Set custom User-Agent header (Mozilla/5.0 (compatible; Bantam/1.0)) to avoid gateway 403 blocks.
   • Retry network/HTTP errors with Fibonacci backoff delays (1s, 1s, 2s, 3s, 5s).
   • If stream=true, parse SSE stream (data: {...}) for real-time reasoning and text output, bracketing reasoning with --- reasoning start --- / --- reasoning end --- markers.
2. Append the assistant's response message object to messages. If non-streaming and response has reasoning tokens (reasoning_content or reasoning), output them wrapped in --- reasoning start --- / --- reasoning end --- markers.
3. If there are pending tool_calls in the assistant response:
   • For each tool call, output a trace log ([tool call: name(args)]).
   • Validate JSON arguments. If invalid, format a tool-error response so the LLM can self-correct.
   • Execute tool action (shell_exec or run_subagent).
   • Output a trace log of the result ([tool result: name]).
   • Append tool result message (role: "tool", tool_call_id, content: result string) to messages.
   • Loop back to step 1.
4. If no pending tool calls (or if max_al_iterations is reached), stop and return messages.

Model configuration parameters

(shared by both implementations; model.cfg is plain key=value with # comments)

• endpoint (base OpenAI-compatible API URL, default https://api.openai.com/v1)
• model (model name, e.g. gpt-4o)
• temperature (model temperature, default 0.7)
• api_key (API key / Bearer token, optional; fall back to OPENAI_API_KEY env var)
• stream (stream response tokens in real-time, default true)
• color (ANSI coloring: auto (TTY-detected, default), always, or never; also disabled by NO_COLOR/BANTAM_NO_COLOR env vars)
• timeout (HTTP timeout in seconds for LLM API calls, default 300; in the Go port it bounds connection setup and time-to-first-byte, so long streaming responses are not cut off mid-stream, matching the Python port's per-operation socket timeout)
• shell_timeout (timeout in seconds for shell_exec commands, default 120)
• max_al_iterations (max tool-call loop iterations per AL() invocation, default 1000)

Both implementations keep the interactive prompt safe against the classic "long line overwrites the prompt" readline bug: the Python port wraps the ANSI escapes in \001/\002 (RL_PROMPT_START_IGNORE/RL_PROMPT_END_IGNORE) markers, and the Go port's built-in editor tracks the cursor with its own column math (terminal auto-wrap aware) and redraws from the first line of the buffer, so wrapped input stays clean at any terminal width.

When enabled, the interactive console uses a subtle ANSI palette: the pending-request status ...requesting... is darkened bold, reasoning markers are cyan, reasoning text is dim, [tool call: ...] traces are yellow, [tool result: ...] headers are green, tool result bodies are dim (red for tool errors/unknown tools), and errors/network retries are red. Tool result payloads fed back to the LLM are never colored.

Tool call definitions

run_subagent tool

• Parameters: prompt (string)
• Return value: string
• Action: run AL(cfg, [{"role": "system", "content": system_prompt + "\n\nImportant: this is a child agent"}, {"role": "user", "content": prompt}]) and return the text content of the last assistant-role message.

shell_exec tool

• Parameters: command (string)
• Return value: string
• Action: run shell command specified in command subject to shell_timeout (default 120s) and return output + '\n\nexit: ' + exit_code string.

Repository layout

• bantam.py — Python reference implementation (stdlib only)
• main.go, term_linux.go, term_darwin.go, term_windows.go, term_other.go — Go implementation (stdlib only, module code.luxferre.top/luxferre/bantam)
• model.cfg, system.txt — shared configuration and system prompt
• README.md — this document

FAQ

Why no MCP support?

If you need MCP server tools, there's nothing a simple shell wrapper cannot solve in this case.

Why no sandboxing?

Same philosophy as the Pi agent: there's nothing a simple chroot environment cannot solve in case sandboxing is really necessary.

Any advanced authentication schemes or header injection?

You can pair Bantam with the Dynagate LLM gateway to achieve all that.

Credits

Created by Luxferre in 2026, released into the public domain with no warranties.