agent-toolkit plugin
agent-toolkit offers generic skills for engineering tasks: from a ticket to a plan, tasks and a handover for review.
- 85 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Installation creates links in agent directories and registers a daily update hook.
- Skills read tickets and pull request comments from connected trackers.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/agent-toolkitDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Installation creates links in agent directories and registers a daily update hook.
- Skills read tickets and pull request comments from connected trackers.
This is third-party code. Review the repository files before installing.
What it does
agent-toolkit is a set of skills for everyday software development, tied to no language or framework. It installs in Claude Code as a plugin and works with any agent that reads the skill format. Skills fall into groups. For docs and skills there are `compact-docs-writer`, `compact-skill-creator` and `self-improve`. For context hygiene there are `agentify-project`, `context-checkup` and `memory-doctor`. The task workflow is a chain: `fetch-ticket`, refinement, `create-implementation-plan`, `harden-artifact`, `split-plan-tasks`, `execute-plan-tasks` and `handover`. Review skills collect pull request comments and go through them one by one.
Who it is for. Engineers who drive tasks with an agent in a refine, plan, act cycle and want a short context in every session.
Good fit when
- A ticket from GitHub, Jira or Azure DevOps must be downloaded as Markdown, refined and turned into a plan.
- The plan is big and must be split into small checkable tasks for a fresh session.
- After the work you need ready text for a pull request description with decisions and a review guide.
- Every session loads too much context and you need to see what can go.
Not a fit when
- The task is small and a chain of ticket, plan and tasks only slows it down.
- You need skills for a specific framework: the set is deliberately generic.
- The team is not ready for a daily auto-update hook in Claude Code.
Example request
Fetch the ticket at this link, refine the requirements, write an implementation plan and split it into tasks for a separate session.Limitations
The `install.sh` script installs skills as symlinks, on Windows the rules are copied and must be updated with `--force` every time. In Claude Code installation registers a hook that updates the set once a day. Rules are installed separately from skills, in a separate README section. Some skills work together: for example `harden-artifact` suggests running `self-improve` when it is installed. To leave a skill out, name it in `./install.sh --exclude` and the choice sticks.
How to disable. Remove the plugin with `/plugin` in Claude Code or the symlinks to skills in agent directories, then remove the daily update hook.
FAQ
How do I install agent-toolkit in Claude Code?
Run `/plugin marketplace add eai-org/agent-toolkit`, then `/plugin install agent-toolkit`.
How do I install all skills in one command?
Run `git clone https://github.com/eai-org/agent-toolkit.git && cd agent-toolkit && ./install.sh`. To update: `git pull && ./install.sh`.
How do I leave out a skill?
Run `./install.sh --exclude use-conversational-language` with your skill name. The choice sticks across daily updates.
What is the RPAC chain?
The refine, plan, act workflow: the ticket is refined, the plan is hardened and split into small tasks, and a fresh session executes them one at a time.
What is memory-doctor for?
It drains the memory agents accumulate on their own, block by block, moving each entry to the right place or archiving it.
Discussion
No questions or reviews yet.
Related
PluginsA library of 30-plus instruction skills for coding agents: planning, building, testing, hallucination guards and shipping
Skills for commits, review and ZenTao plus a status line, gateway balance and session logs for Claude Code, Codex, OpenCode.
Softaworks Agent Toolkit
Softaworks Agent Skills
A marketplace of skills, subagents and commands for coding agents: development, documentation, planning and workplace communication
Builder Skills
Skills for agents
A set of small skills from Builder.io for agent work habits: visual plans and diff recaps, docs-first checks, cross-agent review and clean status updates
A skill that creates and updates AGENTS.md files per the agents.md spec: a thin root file plus separate files for each subsystem
A skills library for Claude Code, Cursor and Codex: team code review, TDD, stacked PRs and repository readiness scoring
README in short
Documentation on the philosophy and installation of the set lives in `docs/core-philosophy.md` and `docs/install-skills.md`. Rules are installed separately.