#417 · Primary category: AI Coding Assistants
tura
Build agent that uses 80% less token and delivers better results.
Project last updated:08/20/26
GitHub Stars
605
Forks
32
Contributors
6
License
AGPL-3.0
Why we included this project
Tura is a terminal-based coding agent built around a simple idea: much of what an agent spends on tokens goes to context it carries through the loop, not to the work itself. Instead of having the model re-enter after every tool result, it compiles the task into a runtime-managed command graph and runs several steps in a single turn. The project publishes benchmark numbers to back that up, reporting 77.5% fewer tokens than Codex CLI on DeepSWE v1.1 tasks with a comparable verifier success rate, and a balanced mode that spends some of the savings on more reasoning and reaches a higher success rate. For developers and small teams whose agent bills are climbing, or who are tired of watching an agent patch a repo before it understands the goal, Tura is worth testing against your own long-horizon tasks.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
superpowers
An agentic skills framework & software development methodology that works.
ECC
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
skills
Skills for Real Engineers. Straight from my .agents directory.
opencode
The open source coding agent.
andrej-karpathy-skills
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.