Bring your X history in
Connect your account through X. PilotMyX imports the posts and replies included in your plan and refreshes recent data daily.
AI agents with CLI access + PilotMyX
Use MCPorter to let a terminal-capable agent inspect your X/Twitter data through PilotMyX's MCP tools. Connect over MCP and ask questions grounded in the posts and replies you actually published.
Which posts earned the most attention last month, and what should I test next?
From account to action
Connect once, give AI agents with CLI access a read-only view of your results, and use the evidence to decide what to try next.
Connect your account through X. PilotMyX imports the posts and replies included in your plan and refreshes recent data daily.
Create a scoped MCP token and install the PilotMyX skill. Ask what performed, over which period, and which real posts support the answer.
Use those findings to draft a new angle. Add separate write permissions only if you want the agent to save or schedule it.
What AI agents with CLI access can do with PilotMyX
PilotMyX supplies the facts. AI agents with CLI access helps you interpret them and plan the next test. It does not need publishing access to analyze your account.
Inspect reach and engagement for individual posts, with the content behind each result.
See whether original posts or replies are doing the work, using a shared time window.
Compare periods and account trends before deciding whether an idea is worth repeating.
Ask for a new angle based on examples from your own account, not generic posting advice.
Grant separate write scopes when you want the agent to prepare or schedule an approved post.
Start read-only, scope each token, and revoke access from PilotMyX when you need to.
The connection
It is the bridge between AI agents with CLI access and your connected X account. Your agent requests cached posts, replies and analytics from PilotMyX, then works with that evidence in your conversation.
PilotMyX does not generate your strategy on the server. Your agent does the reasoning; you decide whether a draft gets published.
Explore the MCP toolsRead-only by default · Publishing requires a separate scope
Try asking AI agents with CLI access
“Compare my original posts and replies from the last 30 days. Which had more reach? Show the sample size and three real examples.”
“Find the posts that beat my usual engagement. What do they have in common? Suggest one angle to test next, but don’t publish it.”
Setup guide
Follow these steps for AI agents with CLI access. Start with a read-only token; add publishing permissions only when you need them.
An agent with shell access can use MCPorter to discover and call PMX's remote MCP tools. MCPorter is a third-party client, not a PilotMyX posting CLI.
Use its installation guide for your operating system. Confirm the executable is available with:
mcporter --version
Run it on the computer where the agent executes commands. Installing it on your laptop will not make it available on a separate cloud computer.
Connect your X/Twitter account to PMX and wait for the import. Create a token with account:read and data:read in MCP Access.
Supply it through the agent computer's private environment as PILOTMYX_MCP_TOKEN. Do not paste it into the agent's conversation, command history or shared files.
Create a private pilotmyx-mcporter.json configuration with this entry. The file contains a variable reference, not the token.
{
"mcpServers": {
"pilotmyx": {
"url": "https://pilotmyx.com/mcp",
"headers": {
"Authorization": "Bearer ${PILOTMYX_MCP_TOKEN}"
}
}
}
}
MCPorter resolves ${PILOTMYX_MCP_TOKEN} when it starts the transport. The configuration reference covers variable syntax and file precedence.
Pass the configuration explicitly so an unrelated client's settings cannot override it:
mcporter --config ./pilotmyx-mcporter.json list pilotmyx --schema
mcporter --config ./pilotmyx-mcporter.json call pilotmyx.account \
--args '{"action":"get","payload":{}}'
Confirm the returned X handle and freshness. Listing tools alone does not prove that the token can read your account.
The compact tool is named account; get is its action. Do not invent commands such as get_account or top_posts.
Install the PilotMyX MCP skill using Copy prompt in MCP Access. It documents the action payloads, date ranges, pagination and scope requirements.
Have the agent compare posts and replies over the same period, then draft a proposed post in its response. Do not enable schedule or publish permissions for an initial analysis.
If the executable is missing, install MCPorter on the agent's actual computer. If the variable is missing, fix that process's private environment. If PMX returns 401, check token validity; if it reports insufficient scope, check the requested action.
Keep command output containing private posts out of public logs. Removing the JSON file does not revoke the token: revoke it in MCP Access as well.
Client settings change. Check the platform's current requirements before connecting.
Read the platform documentation ↗FAQ
No. Start with a read-only token to analyze posts and replies. Drafting, scheduling and publishing require separate write scopes that you choose in MCP Access.
No. PilotMyX provides the account data and publishing tools. AI agents with CLI access does the writing and analysis in your own agent session.
Ask the agent to check the last synchronization time before analyzing a period. PilotMyX refreshes subscribed accounts daily; the MCP skill explains how to check freshness and request a refresh when needed.
Connect your X account, then give AI agents with CLI access the real results behind your ideas.