Dirac: Open-Source Agent Tops TerminalBench with 65.2%, Cheaper and Open
Dirac is an open-source coding agent that just topped the TerminalBench 2.0 leaderboard for gemini-3-flash-preview with a 65.2% score — beating Google's official baseline of 47.6% and the previous top closed-source agent Junie CLI at 64.3%. The run was done fully open-source, with no benchmark-specific AGENTS.md files or other cheating mechanisms. The maintainer submitted a PR to the leaderboard 8 days ago but has not received a response due to backlog.
Key Features
- Hash-anchored parallel edits for efficient and accurate code changes.
- AST manipulation to understand and transform code structurally.
- Context curation to keep context tightly focused, improving accuracy and reducing costs — claims 64.8% average cost reduction vs other agents.
- No MCP (Model Context Protocol) — straightforward tooling.
TerminalBench 2.0 Results
Scored on gemini-3-flash-preview: 65.2% vs Google's 47.6% and Junie CLI's 64.3%. The run was done in a leaderboard-compliant way (no resource or timeout modifications). All code is on GitHub — no difference between what was run and what is public.
Cost Comparison
Dirac's average cost per task across 8 benchmarks (against Cline, Kilo, Ohmypi, Opencode, Pimono, Roo) was $0.18, vs the next best at $0.38. That's a 64.8% reduction, or 2.8x cheaper. For example, Task1 (transformers, 8 files) cost $0.13 vs Cline's $0.37. Task6 (transformers, 25 files) cost $0.34 vs Ohmypi's $0.94.
Installation & Usage
Clone the repo and follow setup instructions in the README.md. The agent runs as a CLI tool. No special setup beyond Node.js and API keys for the chosen model.
📖 Read the full source: HN AI Agents
👀 See Also
Zoku: A Tool That Automatically Detects Repeated Workflows in Claude Code
Zoku is a local tool that hooks into Claude Code's event system to record tool actions across sessions, identifies repeated workflow patterns, and then informs Claude about these patterns so it can proactively suggest or execute them. It requires no configuration, has no dependencies, and stores everything locally in ~/.zoku/.
Claude Review: IntelliJ Plugin for Real-Time Code Review with Claude Code
Claude Review is an open-source IntelliJ plugin that automatically reviews code changes on every file save using Claude Code. It sends unstaged git diffs to Claude with customizable prompts and displays findings as native IntelliJ annotations.
Tilde.run: An Agent Sandbox with a Transactional, Versioned Filesystem
Tilde.run provides isolated, reversible sandboxes for AI agents, with a versioned filesystem that mounts GitHub, S3, and Google Drive, and network isolation by default.
Banker Creates Credit Due Diligence Tool with 31 AI Prompts Using Only Claude
A banker with 17 years of MSME credit underwriting experience in India has open-sourced 31 AI prompts that transform a visiting card into a full credit due diligence report, compressing a 3-4 week process to 30 minutes. The tool was built entirely through conversation with Claude, without writing any code.