Tools

re_gent: Git for AI Coding Agents – Version Control for Agent Activity

✍️ OpenClawRadar📅 Published: May 8, 2026🔗 Source
re_gent: Git for AI Coding Agents – Version Control for Agent Activity
Ad

re_gent is an open-source version control system designed specifically for AI coding agents. It automatically captures every tool call an agent makes — file edits, shell commands, etc. — along with the conversation context that triggered them. Think of it as Git for your agent's actions.

Key Features

  • Automatic tracking: No manual commits. Every tool call is captured as a Step (parent hash, workspace snapshot, conversation delta, tool name and args, session ID, timestamp).
  • rgt log: See what the agent did — file changes, line counts, timestamps, and the causing prompt.
  • rgt blame <file>:<line>: Identify which prompt wrote a specific line, which session, and which tool.
  • rgt rewind (coming soon): Restore workspace to any previous step.
  • Multi-session awareness: rgt sessions lists active sessions; rgt log --session <id> filters by session.
  • rgt show <step-hash>: Full context for a step — diff, conversation, timestamp.
Ad

Installation

# Homebrew (macOS/Linux)
brew tap regent-vcs/tap
brew install regent


Or via Go


go install github.com/regent-vcs/regent/cmd/rgt@latest

Then initialize in your project directory with rgt init. Shell completions for bash, zsh, and fish are included.

How It Works

Agent activity is stored in .regent/ (like .git/):

  • objects/ — content-addressed blobs (BLAKE3)
  • refs/ — session pointers
  • index.db — SQLite query index
  • config.toml

Steps form a directed acyclic graph (DAG). Each session is its own branch; common ancestors are deduplicated. The tool currently supports Claude Code — contributions welcome for other agents.

For the full specification, see POC.md in the repository.

📖 Read the full source: HN AI Agents

Ad

👀 See Also

AI Roundtable: Tool for Comparing 200+ AI Models on Structured Questions
Tools

AI Roundtable: Tool for Comparing 200+ AI Models on Structured Questions

AI Roundtable is a free tool that lets users pose questions with defined answer options, select up to 50 models from a pool of 200+, and get structured responses under identical conditions. It also includes a debate feature where models can see each other's reasoning and a reviewer model that summarizes transcripts.

OpenClawRadar
TextForge: A Claude Code-built email approval tool for LLM workflows
Tools

TextForge: A Claude Code-built email approval tool for LLM workflows

A developer built TextForge using Claude Code to automate email workflows with mandatory approval gates, preventing LLMs from sending emails without explicit permission. The tool integrates with Pipedrive CRM and required Google CASA2 security audit compliance.

OpenClawRadar
Qwen 3.6 27B hits 2.5x speed with MTP speculative decoding on llama.cpp
Tools

Qwen 3.6 27B hits 2.5x speed with MTP speculative decoding on llama.cpp

A Reddit user reports 2.5x faster inference on Qwen 3.6 27B using MTP speculative decoding with a custom llama.cpp PR, achieving 28 tok/s on Mac M2 Max 96GB. Includes pre-converted GGUF quants and fixed chat templates.

OpenClawRadar
nah: A context-aware permission guard for Claude Code
Tools

nah: A context-aware permission guard for Claude Code

nah is a PreToolUse hook that intercepts every tool call in Claude Code, classifying commands by action type like filesystem_read or git_history_rewrite and applying policies based on context. It runs a deterministic classifier in milliseconds with optional LLM escalation for ambiguous cases.

OpenClawRadar