agentleFS
chatmaxima-comchatmaxima-com/chatmaxima-skills/skills/build-chatbot/SKILL.md

build-chatbot

Skill0 starsChanged 8 days agoMITView on GitHub

Build and test a ChatMaxima chatbot or AI agent from a description. Use when the user wants a new bot, an AI agent that answers from their website or documents, a lead capture or booking bot, a FAQ bot, or wants to add steps to an existing bot.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 1 is distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask only what you cannot reasonably assume:

    line 10
  2. Very specific:

    Call list_block_types to see what exists.

    line 21
  3. Very specific:

    Call get_block_schema for every block type you will use, so each field is filled correctly.

    line 25
  4. Very specific:

    Add each block with add_block, passing the previous block's id as after_node so it connects automatically.

    line 39
  5. Very specific:

    Call run_test with a short scripted conversation.

    line 46
  6. Very specific:

    Read the transcript. If a step is wrong, fix it with get_block then update_block (send only the changed fields), and test again.

    line 47

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Build a ChatMaxima chatbot or AI agent

1. Agree the design first

Ask only what you cannot reasonably assume:

  • What the bot is for and who talks to it.
  • Which channel it runs on (website chat, WhatsApp, Instagram and so on). Channels are connected accounts: see them with list_accounts.
  • How a conversation starts: specific keywords, or any message.
  • What it must collect (name, email, phone, date) and where that goes (contacts, a Google Sheet, a calendar booking, a ticket, an email).

Then describe the flow back in a short numbered list and get a yes before building.

2. Pick the blocks

  1. Call list_block_types to see what exists. Common choices:
    • MaxIA AI Agent (ai_prompt): an AI agent with a system prompt, optional knowledge bases, variable extraction and tool calling. Use it when replies should be open-ended.
    • Knowledge Source (knowledge_source): answers one question from the selected knowledge bases and stores the answer in a variable.
    • Text, buttons, questions (ask for name, email, phone, date), conditions, and integration blocks such as Google Sheets, Google Calendar and MaxCRM.
  2. Call get_block_schema for every block type you will use, so each field is filled correctly.
  3. Fill dropdown-style fields from the real lists, never by guessing:
    • knowledge bases: list_knowledge_bases
    • channel accounts: list_accounts
    • integrations and their actions: list_integrations (or list_integration_actions for an app that is not connected yet)
    • calendars: list_calendars; sheet columns: list_sheet_columns
    • contact lists and attributes: list_contact_lists, list_contact_attributes
    • labels, statuses, agents: list_labels, list_statuses, list_team_members
    • voices for a voice agent: list_voice_options

3. Build

  1. create_bot, with welcome_message if the user gave one. The bot starts with a Starting Step (node 7) and a Start Message.
  2. Set the trigger on node 7 with update_block: trigger_type with keywords (exact_match, starts_with, contains, ends_with or any) or allow_any for no keyword. Never leave it unset. Set the inactivity timeout if the user gave one.
  3. Add each block with add_block, passing the previous block's id as after_node so it connects automatically.
  4. For branches, wire each extra output (output_2 and up, for example one per button) to its target with connect_blocks.
  5. In message text, use saved variables with single braces, for example {customer_name}. Never use double braces in bot messages.
  6. Finish with organize_bot, and rename_bot if the name no longer fits.

4. Test before handing over

  1. Call run_test with a short scripted conversation. The first message must match the trigger. Email and SMS are simulated.
  2. Read the transcript. If a step is wrong, fix it with get_block then update_block (send only the changed fields), and test again.
  3. A WhatsApp Flow form cannot be opened in the test chat. Say so, and tell the user to test that part on WhatsApp itself.
  4. Share the transcript summary and the builder link. Tell the user which channel the bot still needs to be connected to, if any.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.