h00man

Plugin h00man in h00mankind/workflow. Commands read from the repo's manifest — today.

Skills (19)

• ambitious

  Turn a rough idea or draft into one ambitious, paste-ready prompt that another agent can run start to finish, in ask mode (the default) or rcmd mode (no questions, recommended guesses). Runs only when the user types /ambitious.

• arena

  Put several agents on one task. The default arena mode runs parallel candidates and grafts the best parts into one result, and benchmark ranks models or skills with a blind judge. Runs only when the user types /arena.

• bruh

  Say it plainly. The default bro mode restates your last message without jargon, eli5 explains a topic in a picture-first HTML page, and caveman switches to terse replies. Runs only when the user types /bruh.

• copy

  Write or revise product and marketing copy. Use for a landing page, a headline, a tagline, a CTA, button labels, UI text, error messages, empty states, tooltips, or onboarding. Skip docs, READMEs, and commit or PR text (writing).

• council

  Get a review or a second opinion on existing work from several strong models of different providers. Runs only when the user types /council.

• fuse

  Combine named upstream skills into one short h00man skill that copies their files through includes.toml, then build and check it. Runs only when the user types /fuse.

• motion

  Add, tune, review, or remove interface motion. Covers animation, transitions, enter and exit, staggers, springs, drag, swipe, sheets, gestures, scroll effects, and janky frames, on the web and in React Native or Expo. Use when the user asks to animate something, "make it feel alive", fix motion that

• name

  Generate and screen brand names for a company, product, or feature, from descriptive to invented, with sound and distinctiveness checks. quick gives 20 names; deep asks questions, then screens 100-plus into an HTML shortlist. Not legal clearance. Runs only when the user types /name.

• plan

  Plan a feature into a PRD with vertical-slice issues in docs/plans, in grill mode or trust mode. Runs only when the user types /plan.

• present

  Build a single-file HTML artifact or a faux product UI to show something. Use for /present, an HTML page, an explainer, a plan, a comparison, a report, a post-mortem, slides or a deck, a diagram, a chart, a demo, a mocked UI, a product shot, or a cursor walkthrough animation. Skip chat answers and s

• prompt

  Write or improve prompts for image, video, and audio generation and editing. Use for an image prompt, an image edit, a video prompt, image-to-video, a voice or TTS prompt, a music or SFX prompt, or a prompt for ChatGPT 2.5, Midjourney, Nano Banana, Grok Imagine, Sora, Runway, Luma, Kling, or Seedanc

• retro

  Look back on agent work and stop its mistakes from coming back. Guard each repeated mistake with a check that fails on its own, keep judgment calls in TASTE.md, and propose environment and skill fixes. Runs only when the user types /retro.

• ship

  Get changes out with reviewer-ready discipline. Runs commit, push, PR, review, babysit, and merge as one loop until the PR lands. Also fixes failing CI, resolves merge conflicts, and rebases or takes over an old PR. Use when the user says ship it, commit this, open a PR, babysit, fix CI, full send,

• teach

  Teach the user a subject, in explain mode (one plain account of code, a change, or a concept, in chat) or course mode (a learning workspace with a mission, lessons, and records over many sessions). Runs only when the user types /teach.

• ui

  Design, build, or review the look of an interface. Sets a visual direction, then works on layout, type, color, spacing, and components. Picks a UI library, stress-tests a screen with worst-case data, makes a web app feel native on a phone, and cleans up UI that looks AI-generated. Use when the user

• variants

  Build 3 to 5 structurally different takes on a component, screen, or flow behind one picker, then recommend one. Runs only when the user types /variants.

• verification

  Prove that behavior works on the real artifact, not a proxy. Use to make or update a project verify or driver skill, also after the app changed ("make a verify skill", "audit the verify skill", "the driver skill is out of date"), to formally verify a race, retry, or invariant ("model-check this", "p

• wdyt

  Give an honest verdict on an idea, plan, or choice, as a senior peer would. Runs only when the user types /wdyt.

• writing

  Write, review, or clean up prose that engineers and agents read. Use for docs, a README, an RFC, an ADR, a guide, commit messages, PR descriptions, a SKILL.md, AGENTS.md or CLAUDE.md, or to unslop, de-AI, or line-edit supplied text. Skip product and marketing copy (copy), verify and driver skills (v

Subagents (4)

• build-agent

  Delegate subagent for code-writing and helper tasks that a chat running the h00man `build` skill spawns. Use it for delegates, not to run `build` itself. Spawn a fresh `build-agent` for each new task. It reads the `build` skill's SKILL.md and the rulebook that SKILL.md names before any work. A `gene

• comment-sicko

  A deranged comment-hater that savors deletion and condemns workaround code.

• thermo-nuclear-code-quality-review-subagent

  Thermo-nuclear code quality audit of a diff: maintainability, structure, file growth past about 1,000 lines, tangled branching, and simpler designs. The h00man `thermos` reference spawns it with the diff and the changed files.

• thermo-nuclear-review-subagent

  Thermo-nuclear branch audit of a diff: bugs, breaking changes, security, devex, and feature-flag leaks. The h00man `thermos` reference spawns it with the diff and the changed files.

Hooks (1)

• SessionStart

  sh "${CLAUDE_PLUGIN_ROOT}/hooks/session-start.sh"

Similar plugins