bridle

An agentic harness for OpenAI-compatible chat completion endpoints.

Features

• Readline-based CLI with tab completion
• Toggle thinking (/think and /no_think)
• Tool calling (read_file, write_file, local_shell, apply_patch) with user confirmation
• Auto load AGENTS.md and CLAUDE.md files
• Oneshot mode (-p)
• Rewind turn history (/rewind and /turns)
• Session management (/save and resume via -r)
• Compaction via context summarization (/compact)

Usage

llama-server --fit on -c 262144 -m /srv/llama/Qwen3.6-35B-A3B-UD-Q4_K_XL.gguf &

./bridle