groniz-cli

The agent skill collection for the Groniz content workflow: plan what to publish, polish the writing, create blog cover images, and publish or schedule the finished content across 32+ social media and chat channels.

This repository includes four skills:

Skill	Purpose
groniz-cli	Publish, schedule, draft, and analyze content with the Groniz CLI.
content-strategy	Build content plans, topic clusters, editorial calendars, and content roadmaps.
humanizer	Edit AI-generated writing so it reads naturally while preserving its meaning and voice.
blog-cover-image	Design and generate raster cover images for blog posts, articles, and social previews.

The Groniz CLI is distributed separately as a self-contained native binary. The other skills do not require the CLI; blog-cover-image requires an available raster image-generation or image-editing capability.

Install the CLI

curl -fsSL https://groniz.com/install.sh | sh

No Node.js required. Update with groniz update (groniz update --check exits 10 when a newer version exists).

Install the skills

npx skills add groniz/groniz-cli

This detects your agent and installs the repository's skills into its skills directory — .claude/skills/ for Claude Code, .agents/skills/ for Cline, Cursor, and OpenCode. To skip the prompt, name the targets:

npx skills add groniz/groniz-cli -a claude-code -a opencode

Install as an Agent Plugin

This repository implements the Agent Plugins 1.0 portable package layout. Point a compatible client at the repository root; the client discovers:

• plugin.json for plugin identity and metadata
• skills/ for the four Agent Skills
• mcp.json for the hosted Streamable HTTP MCP server

Remote MCP authentication is client-managed by the Agent Plugins standard. Use OAuth 2.0 when the client offers it, or configure a Groniz API key as a Bearer token without committing it to this repository. For Codex:

export GRONIZ_API_KEY="your-api-key"
codex mcp add groniz --url https://mcp.groniz.com/mcp \
  --bearer-token-env-var GRONIZ_API_KEY

The .codex-plugin/plugin.json compatibility manifest supplies Codex-specific presentation metadata while the root plugin.json, skills/, and mcp.json remain the portable source of truth.

What the skills cover

Groniz CLI

groniz --help already documents every command and flag, so the skill deliberately does not restate them. It covers what --help cannot:

• Authenticate first — every command fails without credentials, and the stored API key survives across shells and agents where export does not.
• Upload media before posting — raw paths and external URLs are rejected server-side; groniz upload returns the .path to pass back.
• Discover channel settings at runtime — integrations:settings <id> is the source of truth for required settings and character limits, so agents don't hardcode platform assumptions and eat 400s.
• The traps behind the common failures: strict JSON mode shape, differing jq paths, threads via repeated -c, drafts skipping validation, and the missing-release-id flow that leaves analytics unavailable.

Content workflow

• Content strategy — turns business goals, customer research, and competitive gaps into searchable and shareable topics, content pillars, and execution plans.
• Humanizer — detects common AI-writing patterns and rewrites them while preserving facts, intent, structure, and the author's voice.
• Blog cover image — selects an appropriate composition, palette, rendering, typography, mood, and aspect ratio, then creates or revises a raster cover.

Documentation

Public API reference: https://docs.groniz.com/public-api/introduction

MCP Registry

The official MCP Registry metadata for the hosted Streamable HTTP server lives in server.json. The registry entry describes https://mcp.groniz.com/mcp; clients authenticate by supplying their Groniz API key as an Authorization: Bearer ... header.

License

The repository is licensed under Apache 2.0 — see LICENSE. The bundled humanizer skill is licensed under MIT — see skills/humanizer/LICENSE.