neuron_create_flow

Create a new automation flow. A flow is a directed graph of nodes connected by edges. The graph MUST start with at least one trigger node (entry point). Edges connect nodes via output handles. BEFORE BUILDING: call neuron_get_flow_catalog and read its reference — it carries the JSON shapes you cannot guess: the ConditionGroup + operator list for branch/filter/switch conditions, the cases/assignments/weekdays composite shapes, the handle wiring map, and copy-ready examples. Condition field on a logic node must be a {{template}} (e.g. '{{trigger.text}}'); on a trigger's config.conditions it is a raw payload path (e.g. 'text'). AFTER SAVING: the flow always saves, but the response includes a warnings array flagging likely bugs (unknown node type, missing required field, malformed conditions/cases/assignments, edges wired to a handle a node never emits, missing trigger). Always check it and fix any warnings — a warning means that part will silently no-op at run time. TEMPLATE RESOLUTION: Node config values support {{…}} templates resolved at runtime: {{trigger.text}} — the triggering message text {{trigger.contactPhone}} — the sender's phone {{trigger.senderName}} — the sender's name {{vars.myVar}} — a variable set by logic.set_variable {{nodes.n_abc.output.text}} — output from a previous node {{= amount * 1.1 }} — inline expression (arithmetic, comparisons, ternary) ERROR HANDLING: Any node's config can include __errorHandling: { continueOnFail: true, retryCount: 3, retryDelayMs: 1000 } to retry on failure with exponential backoff and/or continue on the 'error' handle instead of failing the run. WIRING RULES: - logic.branch: 'true' and 'false' handles - logic.switch / logic.ai_router: one handle per case id, plus 'default' - logic.wait_for_reply: 'default' (replied) and 'timeout' - logic.wait_for_approval: 'approved', 'rejected', 'timeout' - logic.wait_first_of: 'reply', 'event', 'timeout' - logic.loop: 'loop' (body) and 'done' - action.lookup_contact: 'found' and 'not_found' - action.call_flow: 'default' (returned), 'timeout', 'error' - All other nodes: 'default' handle (or 'error' when continueOnFail is on). It is categorised as a Write tool in the Neuron MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.

neuron_create_flow accepts 4 parameters: name, graph, enabled, description. Required: name. The full parameter table on this page comes from the server's own tool schema.

Register the Neuron MCP server in PolicyLayer and add a rule for neuron_create_flow: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Neuron. Nothing to install.

neuron_create_flow is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.

Yes. Add a rate_limit block to the neuron_create_flow rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.

Set action: deny in the PolicyLayer policy for neuron_create_flow. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.

neuron_create_flow is provided by the Neuron MCP server (neuron-mcp-server). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.