Troubleshooting, FAQ, and your next step

Troubleshooting by symptom

Find your symptom, apply the fix. Most first-week issues are one of these.

Commands do not appear after install. Restart Claude Code. Plugin registration loads at startup, so a restart clears the large majority of day-one problems on its own.

"Agent not found" on a request. The install did not finish. Reinstall with /plugin install marg@marg-marketplace, then restart.

Every report grades MEDIUM. No tools are connected yet, which is the normal starting state rather than a fault. Chapter 24 covers connecting your first one and lifting the grade to HIGH.

A report grades lower than yesterday. A connected tool is unreachable right now. The report names which one, the work continues at the lower grade, and the connection usually recovers on its own.

A long session lost its thread. Open a fresh session in the same folder and run /marg:resume (chapter 7). To prevent the next one, run /marg:compact before a long session so the context is protected in advance.

A note or update seems forgotten. Hard facts about your business belong in the profile, so state them through /marg:btw with a clear signal word. Looser session observations get promoted by /marg:heartbeat, which you can run any time.

Something stalls and you cannot tell why. Run /marg:status first, since it names the blocked items and what each is waiting on. The usual answer is a pending approval gate you have not answered yet.

Questions everyone asks

Where does my business data live? Your profile, memory, and task history are files in your working folder, on your machine or your repo. Connected tools are read through their official APIs using the keys you provide, and chapter 26 governs what is ever allowed to act on them.

Does Marg work without any integrations? Fully. Every team works conversationally from what you describe, deliverables grade MEDIUM instead of HIGH, and you connect tools whenever the upgrade is worth it to you.

Can it send things to my customers? Not on its own. Anything heading outward passes through a human decision, and the approval gates stop money and external changes before they ever fire (chapter 26).

How is this different from a chatbot subscription? A chatbot answers the question in front of it. Marg routes work to specialists who carry context about your business, return graded evidence, track whether their advice actually worked (chapter 22), and compound what gets learned (chapter 23). The gap between the two shows up in month three, not minute three.

What should I do first? Install (chapter 3), onboard honestly, connect the one tool holding the numbers you ask about most (chapter 24), and route a real problem you genuinely have this week.

Your next step

You have seen how the workforce runs. The fastest way to judge it is to look at what it actually produces.

See a real deliverable. The sample 90-Day Growth Blueprint shows the evidence ledger, the binding constraint, and the tracker exactly as a client receives them: appetals.com/sample-blueprint.

Talk to a human. If you want help putting Marg to work on your own company, book a discovery call: appetals.com/book-discovery.