@BohuTANG: After quantification, it turns out that pi's current 4 tools (read / bash / edit / write) plus glob and grep can go up another level. These two are high-frequency tools, but relying on the LLM to piece together a usable bash to replace them is actually quite difficult — it often...
Summary
BohuTANG found through quantitative analysis that adding dedicated grep and glob functions to the pi tool reduces the detours of LLM-generated bash and improves efficiency.
View Cached Full Text
Cached at: 06/08/26, 09:25 AM
After quantization, it was found that pi’s current four tools (read / bash / edit / write), combined with glob and grep, can be taken to another level. These two are high-frequency tools, but it’s quite difficult to rely on the LLM itself to compose useful bash commands to replace them — it often takes detours and repeatedly reads the same file. Giving it dedicated grep/glob tools reduces a lot of detours. https://t.co/lDw3psIyAe
Similar Articles
@XAMTO_AI: The terminal AI coding assistant track is ridiculously competitive. oh-my-pi emerged as a dark horse this week, already surpassing 137k stars. It runs directly in the terminal, and its core selling point is simple—it modifies code with extreme precision. This project is a fork of Pi, with a ton of hardcore additions on top of the original. The most…
Introduces oh-my-pi, a terminal-based AI coding assistant forked from Pi. With its Hashline editing system, it achieves high-precision code modifications, reducing token consumption by 61%. It comes with 32 built-in tools, supports over 40 languages and more than 40 LLM providers, and can import configurations directly from major tools, making migration costs negligible.
@vintcessun: LLMs can now write their own work scripts, decomposing tasks into a group of sub-agents that execute in parallel. A single assistant thinks sequentially, and when faced with tasks like codebase audits or large-scale refactoring, it's either slow or prone to mixing up its thoughts. pi-dynamic-workflows lets the model directly generate a JS script, using agent() and parallel() for task orchestration, runs them in a sandbox, and then aggregates the results—with real-time progress display. The essence is transforming "one person working" into "one person writing a scheduling script, while minions execute in parallel."
Introducing pi-dynamic-workflows, a tool that enables LLMs to dynamically orchestrate multiple sub-agents for parallel task execution by generating JavaScript scripts, suitable for code audits, large-scale refactoring, and similar scenarios.
@geekbb: 为 Pi coding agent 提供纯算法化的会话压缩，无需 LLM 调用
pi-vcc is an open-source tool that provides pure algorithmic conversation compression for the Pi coding agent, achieving 35-99% token reduction without LLM calls, with lossless history search via vcc_recall.
@QingQ77: Run multiple models simultaneously in pi, fuse their responses into one, get better results for less money. https://github.com/leblancfg/pi-fusion… pi-fusion is an extension of pi that adds a "parallel fan-out" workflow to your coding agent.
pi-fusion is an extension of pi that improves performance at lower cost by parallel fan-out of multiple models and fusing results, supporting prompt rewriting and session archiving.
@WWTLitee: Many coding agents easily get stuck when they touch real engineering environments like editors, browsers, and LSP. oh-my-pi takes a different path. Instead of keeping the agent in the terminal chatting, it directly integrates hash anchor editing, LSP, Python, browser, subagents...
oh-my-pi is an open-source coding agent that integrates hash anchor editing, LSP, Python, browser, subagents and other capabilities into a single workspace, providing out-of-the-box complete IDE integration. It currently has 5.8k stars.