Platform Functions

SQL, Python, and AI functions that give agents direct access to world model data and analytics during conversations.

Platform functions are the universal tool primitive for agent data access. You register a SQL query, Python function, or AI composition, and agents can call it during conversations. No container, no deployment pipeline. The function runs on the platform's compute layer and returns results directly to the agent.

Conceptual overview. For background on function types, built-in functions, and how functions relate to Actions and Skills, see the Platform Functions conceptual docs.

Function Types

Type	Description
SQL Table	Parameterized query returning rows across live and analytical data
SQL Scalar	Query returning a single computed value
Python UDF	Custom Python logic on the compute layer
AI Built-in	Composed AI operations (classify, summarize, extract, mask, sentiment)
Foundation Model	Foundation model call with patient context for complex reasoning

Three Tool Categories

Every platform function surfaces as a tool in the agent's context graph. The agent sees a tool name, description, input schema, and result.

Named Functions

Pre-registered functions with fixed input schemas. The agent calls them by name. All named function tool IDs use the fn_ prefix (e.g., fn_caller_history, fn_entity_confidence).

The platform ships with built-in functions covering entity confidence assessment, caller history, patient summaries, intent classification, clinical extraction, PII redaction, care plan generation, handoff summaries, urgency assessment, and memory access. See the conceptual docs for the full list.

Open Query (fn_query)

The agent writes read-only SQL at runtime for questions no named function anticipated. Queries are validated (SELECT/WITH only, DML rejected, row count capped) and run within workspace isolation.

Open Write (fn_write)

The agent records new observations as world model events when no specific write tool exists. Follows the same confidence and write-scope rules as all other world model writes.

Catalog Discovery

Functions can be auto-discovered from the compute catalog. Any function with a description becomes an agent tool automatically at session initialization. Three sources are merged by priority:

1. Catalog discovery - functions found with descriptions (lowest priority)

2. Built-in defaults - platform built-in functions with curated schemas

3. Workspace-registered - functions explicitly registered through the API (highest priority)

Higher-priority sources override lower ones by name.

Context Graph Integration

Functions are available through tool_call_specs on context graph states. If a function is not in the current state's spec, the agent cannot see or call it.

tool_call_specs:
  - tool_id: fn_entity_confidence
  - tool_id: fn_query
    additional_instruction: "Query world model data when no existing tool fits"

Management Endpoints

All endpoints are workspace-scoped under /v1/{workspace_id}/functions.

Operation	Description
List	All registered functions for the workspace
Register	Add a function (must already exist in the compute catalog)
Delete	Remove the workspace registration (underlying function unaffected)
Test	Execute with sample input, returns result with timing
Discover	Query the catalog for available functions with descriptions
Sync	Auto-register all discovered catalog functions not already registered
Query	Execute an open-scope read-only SQL query

CLI

Platform functions can be managed through Agent Forge:

forge platform function list       # List registered functions
forge platform function register   # Register a new function
forge platform function test       # Test with sample input
forge platform function delete     # Remove registration
forge platform function query      # Run an open SQL query
forge platform function catalog    # Discover available functions
forge platform function sync       # Auto-register all discovered functions

Migration from Data-MCP

Platform functions replace Data-MCP for agent data access. The key differences:

	Data-MCP	Platform Functions
Integration	External MCP client required	Built into the agent reasoning pipeline
Access	SQL queries only	SQL + Python + AI operations
Tool resolution	Separate from context graph	State-gated through tool_call_specs
Write capability	Read-only	Read + write (via fn_write)
Discovery	Manual	Auto-discovery from compute catalog

Data-MCP remains available during the transition. New integrations should use platform functions.