Claude AI Not Working? 6 Most-Searched Errors & Fixes (Step by Step)

Claude is now a daily tool for writing, coding and business work — so when it says “at capacity”, stops mid-answer, or gives a confident but wrong fact, it blocks real work. These are the 6 errors users search for most, with the fix for each, step by step.

1. “Claude is at capacity right now”

What you see: Claude refuses new messages, usually in US work hours (9am–6pm Eastern) on the free plan.

Fix — step by step:

1. Wait 10–30 minutes and try again — capacity usually clears on its own.
2. Try off-peak hours: early morning or late evening.
3. Check status.anthropic.com — if there is an incident, it is on Anthropic’s side.
4. If you use Claude for work every day, Claude Pro ($20/month) gets priority access and hits this far less often.

2. “Approaching 5-hour limit” / usage limit reached

What you see: Claude stops mid-session and says you have hit your usage limit.

Fix — step by step:

1. Finish the important task first — the limit resets on a rolling 5-hour window, not at midnight.
2. Do not waste the remaining messages on long retries. Ask Claude to summarise the conversation so far, copy that summary, and paste it into a new chat after the reset.
3. Split heavy work across the day instead of doing it all in one long session.

3. Context window limit / “Your message will exceed the length limit”

What you see: Claude cannot accept a long file, or a long chat starts forgetting earlier instructions.

Fix — step by step:

1. Start a new chat and paste only the part Claude needs for the next task — not the whole history.
2. Split a big job upfront: review one module, one chapter or one file at a time.
3. Use a Project for ongoing work — add your files and instructions there once, and Claude keeps them across chats.
4. Ask Claude to summarise the old chat, copy the summary, and start fresh with it.

4. Claude cuts off mid-sentence / incomplete reply

What you see: The answer stops halfway, usually on a long reply.

Fix — step by step:

1. Just reply “continue” — Claude will pick up where it stopped.
2. End long prompts with: “If your response is cut off, I will ask you to continue.” Then use “continue” when it stops.
3. Break the task into smaller pieces: ask for part 1, then part 2, instead of everything at once.

5. Claude Artifacts not loading / blank screen / send button does nothing

What you see: No error message — just a blank screen, a spinner, or a dead send button.

Fix — step by step:

1. Check status.anthropic.com first.
2. Try the other surface: if the web app fails, open the mobile app (and vice versa) — they fail independently.
3. Open a private window with no extensions, then try a different network. If one works, the problem is local.
4. Send a two-word prompt in a brand-new chat. If that works, the failure is about the size, files or length of the old thread — not availability.

6. Claude gives a wrong fact, refuses, or sounds off

What you see: A confident answer that is wrong, a refusal for a normal task, or the right answer in the wrong tone.

Fix — step by step:

1. For a wrong fact: reply “For each claim in your answer, tell me your confidence (high, medium, low) and where you got it. If you cannot source it, say so.” Claude will usually flag the parts it made up.
2. For a refusal: reply with your real intent in plain English: “You declined that, but here is what I am actually trying to do: [your reason].” Most refusals come from misread intent.
3. For the wrong tone: reply “Rewrite it in the same length but with this tone: [your fix]. Keep all the information, change how it sounds.”
4. Prevention: paste the source instead of asking Claude to remember it, and give one example of the format you want.

Quick trick: the “continue” habit

The single most useful Claude habit is ending every long prompt with “If your response is cut off, I will ask you to continue.” It turns a broken-looking reply into a two-message job.