GodMode plugin
A 33-skill plugin with gates: question the request, verify facts against sources, test first, review by a second model
- 120 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Session hooks inject instructions into every session and prompt
- The cross-model review sends the code diff to a third-party service
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/godmodeDetects 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.
- Session hooks inject instructions into every session and prompt
- The cross-model review sends the code diff to a third-party service
This is third-party code. Review the repository files before installing.
What it does
GodMode turns an agent's work into a chain with gates: until the current stage has evidence, the next one cannot start. First `rationale` checks whether the task is worth doing, `intent-discovery` reads the project, asks questions one at a time and writes a design document, and `grounding` verifies every fact that can change against a current source. Then `task-planning` cuts work into two-to-five-minute tasks with exact files and tests, and `delegated-execution` hands each to a fresh subagent while keeping a ledger on disk in case context is lost. Every task goes test-first. `code-review` checks spec compliance, then quality, and `cross-model-review` sends the diff to a reviewer from another model family: Codex, Gemini or Claude. Before saying "done", `completion-gate` demands fresh command output, and `merge-protocol` lands the branch and cleans up the worktree. The skills are identical on Claude Code, Antigravity, Codex, Cursor and OpenCode. There are two hooks: a session-start bootstrap and a one-line reminder per prompt.
Who it is for. Developers who hand an agent large tasks and want it to check the idea and the facts first and show evidence when it finishes.
Good fit when
- The task is vaguely worded and the agent should first question it and propose two or three approaches
- A feature relies on an API that may have changed and needs a check against current docs, not model memory
- A large plan must run through subagents with a ledger that survives context loss
- Before merging you want an independent review from another model to catch errors the first one misses
Not a fit when
- A one-line edit or rename: gates and two reviews add needless ceremony
- You need a single skill such as `test-first`: the set installs whole, all 33 skills
- You have no subscription or key for a second model family: `cross-model-review` cannot run without it
Example request
Task: add report export to CSV. First question the idea, verify against the docs, then plan and tests.Limitations
The plugin installs a session-start hook that injects a bootstrap and a one-line reminder per prompt, so you need a new session after installing. `cross-model-review` needs access to a second model such as Codex or Gemini, otherwise the step is skipped or done by your own model. Agent-team mode is the most expensive and is used only when tasks must coordinate. On Cursor, OpenCode and other hosts the bootstrap is wired by the host's instructions and behaves differently from Claude Code. `/godmode:doctor` checks the installation.
How to disable. Remove the plugin with /plugin uninstall godmode in Claude Code. For Antigravity, Codex and other hosts use their own removal route described in the README. Skills added with npx skills add live in the agent's skills directory.
FAQ
How do I install GodMode in Claude Code?
Run `/plugin install godmode --marketplace NoobyGains/godmode`. On older versions first `/plugin marketplace add NoobyGains/godmode`, then `/plugin install godmode@godmode`. `/godmode:doctor` confirms the install.
How do I install the skills in Cursor or OpenCode?
The universal route is `npx skills add NoobyGains/godmode`. Automatic triggering also needs the bootstrap route for each host. For OpenCode add `godmode@git+https://github.com/NoobyGains/godmode.git` to the `plugin` array in `opencode.json`.
What does the grounding skill do?
It requires checking any fact that can change, such as a version, an API or a price, against a current source. It forbids relying on model memory when no such source exists.
Does the plugin collect data?
No. Per its README it is Markdown skills plus two small hooks, calls no network service, collects no telemetry and has no runtime dependencies.
How is it different from Superpowers?
It adds evidence gates at each stage, `grounding` fact checks and a mandatory review from another model family. Superpowers gives a basic set of practices without such checks.
Discussion
No questions or reviews yet.
Related
PluginsA skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
A workflow template for Claude Code and five more agents: a backlog task, a diff review and a commit.
A Claude Code and Codex plugin with skills for code research, tests, refactoring and independent review of finished changes
A Claude Code plugin for development with separated roles: spec, review, TDD, QA and verification by different agents
Claude Code kit with eight phase agents, 21 skills and hooks that keep the agent inside set limits.
A Grid Dynamics plugin that loads a team's shared engineering standards, architecture and workflows into a coding agent session
README in short
GodMode is built as a gated pipeline: define, plan, execute, review, ship. It lists all 33 skills by group with each one's prime rule, execution and model routing tables, supported hosts with verification notes, a comparison with other sets and a changelog.