engineering / skill

Cole's AI Skills

Agent skills for a prime, plan, implement, validate, review, commit and PR loop, with worktree skills, hooks and skills for building your own rules.

Repository owner coleam00

Not rated yet
0 comments

Like 0 Save

Published 11 Oct 2026Listing updated 12 Oct 2026

At a glance

Purpose: Agent skills for a prime, plan, implement, validate, review, commit and PR loop, with worktree skills, hooks and skills for building your own rules.
Runs in: Local agent with shell
Operating systems: macOS, Linux, Windows
Needs: Runtime: Not stated; Needs internet: Yes; Needs accounts or API keys beyond your agent: Not stated
Files and data: Reads files: Yes; Writes files: Yes; Sends data to other services: Yes
Tool type: Skill
Delivery: From GitHub
Version: Latest on the main branch
Access: Code on GitHub
Compatibility: According to the README, the plugin route is for Claude Code. The README says the npx skills CLI installs to Codex, Cursor, Copilot, Cline, Windsurf, OpenCode, Continue and other agents. The skills are plain markdown, so the README says most of them work wherever a coding agent can read files. Its drive-screen skill names Windows, macOS and Linux.
External costs: The repository is MIT-licensed and the README states no pricing. Your coding agent or model provider may charge for usage. The README says the skills come from the author's Agentic Coding course and does not state its price.
Permissions: The skills act through your coding agent with the access you give it. According to the README, installation can write project skill files or, with -g, global files under ~/.claude/skills/. The README describes running tests, type checks and lint; creating commits, pushing branches, opening PRs, posting findings to GitHub, writing Jira or GitHub backlogs, creating worktrees and integrating branches. The README describes browser automation and desktop control, including typing, clicking and screenshots. The README says build-signal-engine builds a system that reads changelogs, communities, feeds and videos, filters items and passes selected content to an LLM for a daily digest. The README says rules-create-global derives a root CLAUDE.md from your codebase or specs. The README says hook setup copies Python files into .claude/hooks and copies or merges configuration into .claude/settings.json; hooks run on lifecycle events, can log tool calls and require TEST_COMMAND configuration.
Usage terms: MIT License, as stated in the repository. See its LICENSE file for the exact terms.

Code on GitHub

coleam00/skills

MIT

Python678 stars185 forksLast pushed 7 Oct 2026 UTC

The agent skills I actually use to build software with coding agents. The PIV loop, planning, worktrees, and the meta-skills for building your own AI Layer.

Public GitHub data fetched 13 Oct 2026 UTC · 0 days old

Skill folders in this repository (36)

Folders that contain a SKILL.md file, from GitHub's file list on 13 Oct 2026 UTC.

View on GitHub ↗

What it does

Cole's AI Skills is a collection of agent skills that the README says come straight from the author's .claude/skills folder. The README explains that a skill is a folder with a SKILL.md file holding a name, a description of when to use it, and the procedure the agent follows. It says the skills are built around one loop: prime, plan, implement, validate, review, commit, PR.

The README groups the skills by purpose: priming a codebase, backend or frontend; planning PRDs, architecture, epics and stories; the piv skills for implementation plans, implementation, a validation run, change review, commits and pull requests; investigating and fixing GitHub issues; creating and merging git worktrees; and meta-skills for rules, hooks, skills and process review. It also lists build-dark-factory, build-signal-engine, and tool skills for browser automation, AST code search and desktop control.

A hooks folder holds copy-in hooks with their own README. The README says they block routes to secrets, refuse rm -rf, log every tool call and refuse to let the agent finish while the tests are red, among others.

Who it is for

Developers who build software with a coding agent such as Claude Code and want ready-made skills and hooks for planning, implementing, reviewing and shipping changes.

Example tasks

Orient the agent in a codebase before planning, with prime-codebase.
Interview your way to a problem-first PRD with plan-create-prd.
Run piv-validate, which the README says runs the project's full suite and returns one PASS or FAIL verdict.
Create several git worktrees for parallel work, then merge them with worktree-merge.
Describe a rule such as never touching migrations to hooks-create to get a working hook.

Getting started

The README gives four routes. The README says to run /plugin marketplace add coleam00/skills and then /plugin install skills@cole-medin inside Claude Code. Plugin skills use namespaced commands such as /skills:piv-implement. As editable files for any agent, run npx skills add coleam00/skills, optionally with -g to install globally. You can also clone the repository and copy a skill folder into your project's .claude/skills/ folder, or ask your agent to do that. For those last two routes, restart your session or run /skills. The README has the exact commands.

Limitations

The README says the plugin route is for Claude Code and that the npx skills CLI installs the skills for other agents, naming Codex, Cursor, Copilot, Cline, Windsurf, OpenCode and Continue. Skills that call each other with slash-command syntax assume Claude Code's invocation; elsewhere the README says to write "use the piv-validate skill". The piv-validate skill ships with a placeholder command list that you replace with your project's test, type-check and lint commands. The README says piv-commit and piv-create-pr read .claude/references/conventions.md if it exists. The setup-ai-tutor skill is sample-specific, so the README says to adapt or delete it. The README lists Windows, macOS and Linux for drive-screen. Hooks do something as soon as they are installed. The README says to read them first and set TEST_COMMAND in the stop hook. The plan-create-stories skill targets Jira or GitHub. The piv-review-pr skill hands its deep pass to a code-reviewer subagent if one is in .claude/agents/, and otherwise reviews inline. The repository can change at any time; check its README before you install.

Data practices

The README has no data-handling or telemetry section. It says the piv-review-pr skill posts its findings to GitHub, the plan-create-stories skill writes a backlog to Jira or GitHub, and one hook logs every tool call. What is sent where otherwise depends on your coding agent and model provider.