Skip to main content

INTEGRATION SURFACES

Developer Access

instr-core follows a deliberate sequence: plan → validate → explicit confirmation → execute. Planning results must first pass Schema, state, range, and dry-run validation; only after explicit operator confirmation can the execution layer proceed.

instr-core on GitHub

CONTROLLED WORKFLOW

Clear human confirmation points between planning and execution

  1. 01

    Plan

    Generate inspectable operation plans based on task and instrument Schema.

  2. 02

    Validate

    Validate identity, capability, range, state, prerequisites, and dry-run.

  3. 03

    Confirm

    Present intended actions and risks to the operator for explicit confirmation.

  4. 04

    Execute

    Only after confirmation hand constrained actions to the local connection and driver layer.

EXECUTION SURFACES

From Agent to local instrument connection

MCP

Expose instrument knowledge, Schema, and constrained tool capabilities to compatible AI clients.

Agent API

Provide explicit, auditable planning and execution interfaces for applications and agents, instead of implicit automation.

Desktop

Host connection, confirmation, state feedback, and execution records in the operator's desktop environment.

VISA / SCPI

The local instr-core execution layer connects to instruments and checks model, state, range, and prerequisites before sending commands.

DATA MODELS

Resolve stable data models before planning

Data models use the same instrument_key as instr-core to publish human-readable summaries of identity, capabilities, ranges, prerequisites, state transitions, and safety limits. Agents should resolve and validate the schema before generating dry-runs and pending confirmation plans.

Browse data models