tuffest harness

This is a tiny AI harness, not just a subprocess wrapper.

prompt.md is the model prompt seed. harness.py loads an OpenAI-compatible client, calls the model, and exposes five tools:

• shell (command execution)
• list_dir (read-only directory listing)
• read_file (read-only file window)
• git_status (read-only git status)
• harness_state (self-state summary)

main.py is presentation + CLI flow: transcript rendering, rich/plain output mode, and optional prompt watching.

runtime

• Set CEREBRAS_BASE_URL, CEREBRAS_API_KEY, and CEREBRAS_MODEL_STRING in .env.
• Run one turn with uv run main.py --run-once.
• Run one turn with a seed message: uv run main.py --run-once "wsp".
• Start interactive chat loop with uv run main.py.
• Watch prompt.md while chatting with uv run main.py --watch.
• Watch-only (no stdin chat) with uv run main.py --watch --no-chat.
• Disable rich rendering with uv run main.py --rich no.

chat controls

• Type a message and press Enter to append a user message and run a turn.
• Use /run to force a turn without adding a user message.
• Use /exit or /quit to stop.

session

The session file is append-only JSONL. Each event is either a model message, a tool result, or an error event. That keeps the loop dumb and easy to inspect.

shape

• tools_runtime.py: tool schemas and tool handlers.
• harness.py: model client, turn orchestration, transcript append.
• transcript_view.py: transcript rendering and rich/plain output formatting.
• main.py: CLI, chat loop, and watch loop wiring.
• prompt.md: the behavior seed for the model.