neuron

Create Bot

neuron_create_bot

Create a new AI bot with specified name, model, and behavior instructions. Returns the created bot object.

Input Schema

Name	Required	Description
`name`	Yes	Display name for the bot
`llmModel`	No	OpenRouter model ID (e.g., 'deepseek/deepseek-v4.1-flash' (default), 'google/gemini-2.5-flash')
`metadata`	No	Arbitrary metadata object stored with the bot. Known keys: 'communityExitFollowUp' (auto-DM users who leave watched groups — { enabled: boolean, groupJids: string[], message: string with {name} placeholder }), 'learningConfig' (self-learning from conversations — { enabled: boolean, folder: string, groupChats: string[], dmPhones: string[], useAdminPhones: boolean }), 'loadBalancing' (multi-channel round-robin — { enabled: boolean, strategy: string }), 'welcomeNewMembers' (fine-tune welcome behavior — { batchWindowSeconds: number, groupJids: string[] }). Additional custom keys are preserved as-is.
`maxTokens`	No	Maximum number of tokens the bot can generate per response
`assignment`	No	One-sentence role definition for the bot (max 2000 chars)
`systemPrompt`	Yes	System prompt that defines the bot's behavior, personality, and response guidelines
`throttleMode`	No	While paused: 'hold' (default) queues sends and delivers them when safe — the bot API returns success with throttle.status 'held'; 'drop' refuses them — the bot API returns 429 THROTTLED with Retry-After.
`llmTemperature`	No	Temperature parameter controlling response randomness (0 = deterministic, 2 = creative)
`welcomeMessage`	No	Welcome message sent to new members joining groups the bot manages. Setting this field activates the batched welcome feature — new members are accumulated over a window (default 60s) and welcomed in a single message with @mentions. Set to null to disable.
`fallbackMessage`	No	Message sent when the bot cannot understand user input (max 2000 chars)
`greetingMessage`	No	Automatic greeting sent when a new conversation starts (max 2000 chars)
`throttleEnabled`	No	"Protect my number": slow down, then pause, the bot's proactive WhatsApp sends (API, flows, tasks) when the number nears volume that gets numbers suspended, and resume automatically. Replies to inbound messages are never held. WhatsApp doesn't publish its limits: these are conservative estimates that cut the risk but can't guarantee no suspension; content and spam reports matter as much as volume.
`escalationPrompt`	No	Prompt template used when escalating to a human agent (max 5000 chars)
`responsibilities`	No	Array of responsibility descriptions defining what the bot handles
`throttleHoldHours`	No	Hold mode: give up on a held message after this many hours (default 24); it then appears under failed messages.
`whatsappChannelId`	No	Unique identifier (UUID) of the WhatsApp channel to associate with the bot

Schema Changelog

Changes observed during successful MCP inspections.

Changed3 schema fields changedOct 16, 2026

addedInput schema / properties / throttleEnabled

Added value: +{
+  "description": "\"Protect my number\": slow down, then pause, the bot's proactive WhatsApp sends (API, flows, tasks) when the number nears volume that gets numbers suspended, and resume automatically. Replies to inbound messages are never held. WhatsApp doesn't publish its limits: these are conservative estimates that cut the risk but can't guarantee no suspension; content and spam reports matter as much as volume.",
+  "type": "boolean"
+}

addedInput schema / properties / throttleHoldHours

Added value: +{
+  "description": "Hold mode: give up on a held message after this many hours (default 24); it then appears under failed messages.",
+  "maximum": 72,
+  "minimum": 1,
+  "type": "integer"
+}

addedInput schema / properties / throttleMode

Added value: +{
+  "description": "While paused: 'hold' (default) queues sends and delivers them when safe — the bot API returns success with throttle.status 'held'; 'drop' refuses them — the bot API returns 429 THROTTLED with Retry-After.",
+  "enum": [
+    "hold",
+    "drop"
+  ],
+  "type": "string"
+}

Changed1 schema field changedOct 7, 2026

changedInput schema / properties / llmModel / description

Previous value: -"OpenRouter model ID (e.g., 'google/gemini-2.5-flash-lite', 'openrouter/auto')"New value: +"OpenRouter model ID (e.g., 'deepseek/deepseek-v4.1-flash' (default), 'google/gemini-2.5-flash')"

Changed2 schema fields changedJul 25, 2026

changedInput schema / properties / metadata / description

Previous value: -"Arbitrary metadata object stored with the bot. Known keys: 'communityExitFollowUp' (auto-DM users who leave watched groups — { enabled: boolean, groupJids: string[], message: string with {name} placeholder }), 'learningConfig' (self-learning from conversations — { enabled: boolean, folder: string, groupChats: string[], dmPhones: string[], useAdminPhones: boolean }), 'loadBalancing' (multi-channel round-robin — { enabled: boolean, strategy: string }). Additional custom keys are preserved as-is."New value: +"Arbitrary metadata object stored with the bot. Known keys: 'communityExitFollowUp' (auto-DM users who leave watched groups — { enabled: boolean, groupJids: string[], message: string with {name} placeholder }), 'learningConfig' (self-learning from conversations — { enabled: boolean, folder: string, groupChats: string[], dmPhones: string[], useAdminPhones: boolean }), 'loadBalancing' (multi-channel round-robin — { enabled: boolean, strategy: string }), 'welcomeNewMembers' (fine-tune welcome behavior — { batchWindowSeconds: number, groupJids: string[] }). Additional custom keys are preserved as-is."

addedInput schema / properties / welcomeMessage

Added value: +{
+  "description": "Welcome message sent to new members joining groups the bot manages. Setting this field activates the batched welcome feature — new members are accumulated over a window (default 60s) and welcomed in a single message with @mentions. Set to null to disable.",
+  "type": "string"
+}

Changed1 schema field changedJul 22, 2026

addedInput schema / properties / metadata

Added value: +{
+  "additionalProperties": {},
+  "description": "Arbitrary metadata object stored with the bot. Known keys: 'communityExitFollowUp' (auto-DM users who leave watched groups — { enabled: boolean, groupJids: string[], message: string with {name} placeholder }), 'learningConfig' (self-learning from conversations — { enabled: boolean, folder: string, groupChats: string[], dmPhones: string[], useAdminPhones: boolean }), 'loadBalancing' (multi-channel round-robin — { enabled: boolean, strategy: string }). Additional custom keys are preserved as-is.",
+  "type": "object"
+}

Changed1 schema field changedJul 9, 2026

changedInput schema / properties / llmModel / description

Previous value: -"OpenRouter model ID (e.g., 'google/gemini-2.5-flash-lite-preview-09-2025', 'openrouter/auto')"New value: +"OpenRouter model ID (e.g., 'google/gemini-2.5-flash-lite', 'openrouter/auto')"

First observedJun 28, 2026

TDQS

B3.2/5.0

Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare the mutation profile (readOnlyHint=false, idempotentHint=false, destructiveHint=false), so the safety burden is covered. The description usefully notes it returns the created bot object (no output schema exists), but omits auth/permission needs, validation behavior, and the side effects of behavior-heavy options like throttleEnabled or welcomeMessage. With annotations carrying safety, this is adequate but thin.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Two short sentences with zero waste and the core action front-loaded. No filler or redundancy, though it is extremely terse for a 16-parameter tool.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 16-param tool with nested objects and one enum but no output schema, the description is sparse: it does note the return object, and the schema/annotations are rich, but it says nothing about required-vs-optional fields or the many behavior-configuring options. Adequate given structured coverage, but leaves gaps an agent would benefit from having filled.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the schema fully documents all 16 parameters, including the rich nested metadata object. The description only gestures at 'name, model, and behavior instructions,' adding no syntax, defaults, or constraints beyond what the schema already provides. Baseline 3 is appropriate.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb ('Create') and resource ('AI bot') and lists the key inputs (name, model, behavior instructions). It is clearly distinguishable from neuron_update_bot, neuron_delete_bot and neuron_get_bot by the verb, but does not explicitly name or contrast with those siblings.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No when-to-use guidance, no prerequisites, and no alternatives cited. It never says when to create a bot versus updating one or vs. neuron_apply_persona_to_bot, leaving all routing to inference from the name.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All neuron tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources

Need Help?