A developer released cli-faq-shortcuts, an Agent Skill that scans your Claude Code and Codex prompt history, finds frequently repeated requests, and turns them into short commands like /sent. It works across Claude Code, Codex and Cursor.
A developer tested Activeloop's Hivemind, a continual-learning shared-memory layer for coding agents, in a real workflow for a day. It extracted 19 named skills from agent history in about two minutes, each with provenance, and refused to create a skill for a workflow already documented in the repo. The tool demonstrates selective memory, avoiding redundancy.
A developer tested Activeloop's Hivemind, a continual-learning shared-memory layer for coding agents, in a real workflow for a day. It extracted 19 named skills from agent history in about two minutes, each with provenance, and refused to create a skill for a workflow already documented in the repo. The tool demonstrates selective memory, avoiding redundancy.
SkillMine is a Go CLI that analyzes your local Claude Code transcripts to identify prompts you repeat and the commands Claude runs in response, then helps you turn those into reusable skills. It uses TF-IDF and cosine similarity over raw JSONL files, fully local with no index, and includes a skill so the agent can run it itself. It solves the problem of manually crafting skills from repetitive prompts for Claude Code users.
SkillMine is a Go CLI that analyzes your local Claude Code transcripts to identify prompts you repeat and the commands Claude runs in response, then helps you turn those into reusable skills. It uses TF-IDF and cosine similarity over raw JSONL files, fully local with no index, and includes a skill so the agent can run it itself. It solves the problem of manually crafting skills from repetitive prompts for Claude Code users.
A developer asks whether to switch from Cursor to terminal-based AI coding, comparing Claude Code against OpenAI's Codex paired with the Pi harness. The discussion centers on model flexibility, self-hosting options, and whether Claude Code's limitations justify its use.
A developer asks whether to switch from Cursor to terminal-based AI coding, comparing Claude Code against OpenAI's Codex paired with the Pi harness. The discussion centers on model flexibility, self-hosting options, and whether Claude Code's limitations justify its use.
RUC DataLab released SkillAdam, an open-source tool (MIT license, Python 3.10+) that automatically improves AI agent SKILL.md files. It tests a skill on relevant tasks, learns from results and prior revision feedback, and applies only changes that pass evaluation. It ships as integrations for Codex, Claude Code, Cursor Agent, and GitHub Copilot, letting developers optimize skills inside their existing agent.
RUC DataLab released SkillAdam, an open-source tool (MIT license, Python 3.10+) that automatically improves AI agent SKILL.md files. It tests a skill on relevant tasks, learns from results and prior revision feedback, and applies only changes that pass evaluation. It ships as integrations for Codex, Claude Code, Cursor Agent, and GitHub Copilot, letting developers optimize skills inside their existing agent.
A developer built OpenHub, a terminal-based tool that indexes and installs AI coding agent skills, prompts, and MCP servers from a central repository. It exports skills as SKILL.md files and handles binary/MCP server setup, aiming to eliminate manual configuration.
A developer built OpenHub, a terminal-based tool that indexes and installs AI coding agent skills, prompts, and MCP servers from a central repository. It exports skills as SKILL.md files and handles binary/MCP server setup, aiming to eliminate manual configuration.