SKILL.md file that teaches an AI agent to use the PostQueen CLI. An agent that has read it lists your channels first, reads what each network needs, uploads media before it attaches it, and passes the settings each post needs. Without it, an agent tends to guess at flags and send posts the API refuses.
The skill is for agents that run shell commands, such as Claude Code, Cursor, Gemini CLI, Grok Build, OpenClaw and Hermes Agent. Chat assistants such as Claude, ChatGPT and Grok do not run commands on your machine: connect them over MCP instead.
What it teaches
Rule 1: authenticate first
Check
postqueen auth:status before anything else. Every command fails without a valid key.Rule 2: upload media first
Every file passed to
-m goes through postqueen upload. A file on disk or a link to another site is refused.Rule 3: TikTok publishes with DIRECT_POST
UPLOAD only drops the video into the TikTok app’s inbox. The skill uses DIRECT_POST unless you ask to finish the post in TikTok.Rule 4: read the settings first
postqueen integrations:settings says what a network needs. A setting that does not apply is dropped without an error.- The core workflow: authenticate, discover channels, look up values, upload, post, check analytics, and fix a post whose published copy PostQueen could not find.
- Which networks connect today: the ones that work, the ones limited until the network approves PostQueen, and the ones marked Soon, so the agent does not promise a network you cannot connect.
- Settings by network, with examples that pass PostQueen’s checks, and the helpers that look values up.
- How to read the output: one status line, then JSON, so the agent drops the first line before it parses.
Before you start
- The CLI where the agent runs:
npm install -g postqueen, with Node.js 20.19 or newer. On Node.js 22, use 22.12 or newer. The skill tells the agent to install it when it is missing, but installing it first saves a step. - The API key as
POSTQUEEN_API_KEY, in the environment the agent’s commands run in. The skill carries no key.
Where the API key is: in the app, open Connections > API Keys. Only a workspace Admin or Super Admin can reveal or rotate it; other members see it masked. A workspace has one key, and each agent card on Connections shows your MCP address with the key already filled in.
Install the skill
- Any agent
- Claude Code plugin
- Grok Build plugin
The It asks which agents should get it and installs it in the current project. Add
skills command reads the skill from GitHub and installs it for the agents on your machine:-g to install it for your user in every project, and -a to name the agent:Most agents load skills when a session starts, so start a new one after installing.
Give the agent the key
The agent runspostqueen in its own environment, which is not always your terminal.
- A coding agent in your terminal reads the key from your shell. Add
export POSTQUEEN_API_KEY=your_api_keyto your shell profile, such as~/.zshrc, and open a new terminal. - OpenClaw runs as a background service that does not see your terminal. Put the key under
skills.entries.postqueen.envinopenclaw.json, or in the service’s own environment, and restart it. See OpenClaw. - Codex asks before every command that uses the network, so unattended runs stop. Connect Codex over MCP instead.
Check that it loaded
Start a new session and ask:postqueen integrations:list and shows your channels. To check the key on its own, run postqueen auth:status: it names the key it uses and says Credentials are valid when the API accepts it.
More fixes: troubleshooting agents.
Keep it up to date
/plugin marketplace update postqueen-agent fetches the latest version and /plugin uninstall postqueen@postqueen-agent removes it.
Next steps
The CLI
Install
postqueen, set the key and schedule a first post.Command reference
Every command and flag the skill uses.
Examples by network
Commands with the settings each network needs.
Every agent
Which agents use the skill and which connect over MCP.