Agent skills are small instruction files that live inside your project and tell a coding agent how to use a CLI tool correctly. The Groniz skill teaches Claude Code, Codex, Cline, Cursor, opencode, ChatGPT, and other skill-aware agents how to authenticate, upload media, discover channel settings at runtime, and avoid the traps thatDocumentation Index
Fetch the complete documentation index at: https://mintlify.com/groniz/groniz-cli/llms.txt
Use this file to discover all available pages before exploring further.
groniz --help alone cannot warn against — so your agent can schedule posts across 28+ channels without guesswork or avoidable failures.
Install in one command
Run this from the root of any project where you want agent support:skills CLI detects which agent you are using and installs the skill file into the right directory automatically. To skip the detection prompt and target specific agents, name them with the -a flag:
Where the skill file is installed
| Agent | Install location |
|---|---|
| Claude Code | .claude/skills/ |
| Cline | .agents/skills/ |
| Cursor | .agents/skills/ |
| opencode | .agents/skills/ |
| Codex | .agents/skills/ |
| ChatGPT | .agents/skills/ |
What agents learn
The skill covers whatgroniz --help cannot: the four things every agent needs to know before it can post reliably.
Authenticate first
groniz auth login stores your API token in ~/.groniz/credentials.json. This credential survives across shells and agent restarts, unlike export, which disappears when a session ends. Every command fails without valid credentials, so the skill instructs agents to verify authentication before attempting anything else.Upload media before posting
Raw file paths and external URLs are rejected server-side. The skill tells agents to always run
groniz upload first and pass the returned .path value to the post command. Skipping this step produces a 400 error that is not obvious from --help alone.Discover channel settings at runtime
Required fields and character limits vary by integration and account configuration. The skill instructs agents to call
groniz integrations settings <id> to retrieve the source-of-truth settings for any channel before composing a post, rather than hardcoding platform assumptions.Avoid common traps
The skill documents five failure patterns that
--help does not explain: the strict JSON mode shape required by the --json flag, differing jq paths across output types, thread chaining via repeated -c flags, draft behavior that skips server-side validation, and the missing release ID that leaves post analytics unavailable.Supported agents
Claude Code
Install via skill or MCP. The skill drops into
.claude/skills/ and is read automatically. MCP is also supported with a paid Claude plan.Codex
Install via skill into
.agents/skills/, or connect via MCP using ~/.codex/config.toml. The GRONIZ_API_KEY env var must be set before Codex launches.Cline
Install via skill into
.agents/skills/. Cline reads that directory automatically and also supports the streamable HTTP MCP endpoint.Cursor
Install via skill into
.agents/skills/. Cursor reads that directory automatically — no additional configuration needed.opencode
Install via skill — opencode reads
.claude/skills/ as well as .agents/skills/. MCP is also supported via opencode.json.ChatGPT
Install via skill into
.agents/skills/. Available on Business, Enterprise, and Edu plans only — not Free, Plus, or Pro.The Groniz CLI skill is open source at github.com/groniz/groniz-cli under the Apache 2.0 license. You can read the raw
SKILL.md, fork it, or contribute improvements.