Skip to content

What FixMyPrompt does

Check and fix a prompt before you send it, then let your AI keep what it worked out so it never starts from zero.

Fix any prompt in seconds

Paste a prompt and get a score out of 100, a rewrite, and the reasons. It also tells you when a prompt cannot run at all.

  • Score and rewrite

    Every prompt gets a score out of 100, a rewritten version and a short list of what changed.

  • Checks by type

    A question, a system prompt, an agent task and a workflow are each checked against what that kind of prompt needs.

  • Flags prompts that cannot run

    When the AI tool cannot do the job no matter how it is worded, the check says so instead of polishing it.

  • Tighten

    Cuts the tokens in a saved prompt and shows the real count before and after.

Check my prompt

Coming

Planned and not available yet.

  • What a prompt costsComing

    See what a prompt costs to run on each AI, and how much Tighten saves.

Fix the files your AI agent reads

Your coding agent reads a CLAUDE.md or AGENTS.md at the start of every session. The doctor shows what that file costs, what it is missing and where its rules clash, then fixes it.

  • Check an agent file

    Paste a CLAUDE.md or AGENTS.md and see the tokens it costs every session, the rules it is missing and the blocks an installer pasted in.

  • Full Agent DoctorPro

    Checks several agent files together for rules that clash, asks you for the facts it cannot see, and hands back a fixed file with every change listed.

  • Agent Doctor in Claude CodePro

    Run the doctor command to check the agent files in your project without sending your source code, then write the fix with your own edit permissions.

Check an agent file

Coming

Planned and not available yet.

  • A check on every pull requestComing

    Prompt and agent files are checked again each time a pull request changes them.

  • Memories written into your agent fileComing

    The memories you use most are kept in your CLAUDE.md for you.

Your AI never solves the same problem twice

When your AI finally works something out, it saves how: the steps, how to check it worked, and what not to do. Next time it follows them.

  • Try memories in the app

    Write and keep memories in the app to see how they work before you connect a tool.

  • Saves what your AI figured outPro

    Say "save how you fixed that to memory" and the steps, the check and the mistakes to avoid are kept.

  • Same memory in every AI toolPro

    Connect your AI tools once and each of them finds the same memories.

  • Run it as a slash commandPro

    Give a memory a short name and run it as a command in Claude Code.

  • Steps with blanksPro

    Leave a blank in the steps and your AI asks you to fill it each time it runs them.

  • Knows when steps stop workingPro

    Your AI reports whether the steps worked, and a memory that failed is marked so it gets fixed.

  • Versions and undoPro

    Every change is a new version with a mark on the one in use, and you can go back to an earlier one.

  • MergePro

    Two memories that cover the same job become one.

  • Secret check before saving

    Before anything is stored, it is checked for passwords, keys and other secrets.

  • Saves lessons as you workPro

    In Claude Code, when your AI gets something wrong and then finds what works, it saves the lesson on its own as a draft. You can turn this off.

  • Import past sessionsPro

    Run the import command in Claude Code and your AI picks the lessons from past sessions on your machine. Only the lessons are sent, never your code.

  • Import from AI chat appsPro

    Copy a ready-made prompt into an AI chat app and paste back the lessons it finds, or upload a chat export and see the credit cost before it is read.

  • Drafts you look overPro

    Every saved or imported lesson is marked as needing a look, so you keep, edit or delete it, one at a time or all at once.

  • Your own AI rulesPro

    A short page of your own rules, loaded into every session you start.

  • Session note in Claude CodePro

    Claude keeps a short private note as you work, and the plugin puts it back after the conversation is compacted.

How memories work

Coming

Planned and not available yet.

  • Document uploadComing

    Upload a document and get memories built from it.

One person's fix helps the whole company

One person saves how a job is done and the rest of the company finds it, with an admin approving changes to what is shared. Enterprise adds project leads and custom limits.

  • Shared with your companyBiz

    A memory one person saves can be found by a teammate's AI in their own session.

  • Admin approvalBiz

    Changes to shared memories wait for an admin to approve them.

  • Teams and adminsBiz

    Group people into teams and choose who runs the company account.

  • Choose what gets sharedBiz

    Share everything saved in a team project, or let each person choose what to share.

  • Know-how stays when people leaveBiz

    When someone leaves, their seat is freed and what they saved stays with the company.

  • Credits shared by the companyBiz

    Each seat adds credits to one company pool that everyone draws from.

  • Project leadsEnterprise

    Put a lead on each project to approve changes, manage who has access and set its rules.

  • Stricter rules per projectEnterprise

    A lead can make one project stricter than the company default, never looser.

  • One review pageEnterprise

    Pending changes, memories that failed last time, recent shares and who has access, across projects, in one list.

  • Custom seats, limits and billingEnterprise

    Seats, memory limits and billing are set up with you, with no self-serve checkout.

  • Company and role pagesBiz

    One page of company rules, and a page per job role if you want one, loaded into every session. Lines you lock cannot be overridden.

  • New hire onboardingBiz

    A new person's role page tells their AI where things are, what not to touch and who to ask, from their first session.

  • Company rules in every checkBiz

    A prompt check run by someone in your company flags a prompt that goes against a locked rule, and the suggested fix follows it.

  • Savings estimate and quality reportBiz

    An estimate of the tokens and time Team Brain saved, with the math shown, plus what changed on your pages, who approved it and which clashes were caught.

  • A key per personBiz

    Each person can use their own AI key, so their AI use is billed to that key. Admins see who has a key set, never the key.

See plans

Coming

Planned and not available yet.

  • Company documents as memoriesComing

    Standards, specs and brand guides become memories linked to the section they came from.

  • Ask from Slack or Microsoft TeamsComing

    Ask the team memory a question from the chat tool you already use.

  • Single sign-on and user provisioningComing

    Sign in through your identity provider, and people added or removed there are added or removed here.

  • Audit log exportComing

    Export the record of who changed what and when.

Ready-made prompts for the work you do

Ready-made prompts sorted by industry and job. Each one already passes the check, and you can open any of them in the checker to adapt it.

  • Prompts for insurance agents

    Client emails, quotes, renewals, follow-up and more, sorted by the job you are doing.

  • Open in the checker

    Send any library prompt to the check to fit it to your own situation and see its score.

  • Personal libraryPro

    Answer a few questions about your business and your prompts come with your details filled in. Prompts we write for your business pass our check before they reach you.

  • Your go-to promptsPro

    The prompts that work for you move to the top, one click to copy, and they are saved to your memories so every connected tool has them.

Browse the prompt library

Coming

Planned and not available yet.

  • More industriesComing

    Real estate, mortgage, legal and dental are next.