Configuring Block Policy

SkillAI & models

Guide users through creating and managing .satk/block-policy.json for controlling which blocks the agent can use, which are excluded, and which block parameters the agent should not modify. Use when setting up block usage policy for a project.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Configuring Block Policy skill

What this skill tells your AI

The instructions your AI receives, as published by matlab/simulink-agentic-toolkit in skills-catalog/model-based-design-core/configuring-block-policy/SKILL.md and read by ahel’s review.

Create or update .satk/block-policy.json — the file that controls which blocks (library or built-in) the agent can use, which are excluded, and which block parameters the agent should not modify.

When to Use

  • User asks to set up, create, or configure block policy
  • User asks to restrict or exclude blocks
  • User wants to prevent the agent from modifying certain block parameters
  • building-simulink-models routes here via the policy gate

When NOT to Use

  • Actively building a model → building-simulink-models
  • Improving block descriptions or categories → simulink-curating-library-kg
  • Declaring which libraries exist → Library Setup gate in building-simulink-models

Prerequisites

  • .satk/reuse-libraries.json must exist.

Workflow

  1. Check state — Does .satk/block-policy.json exist? If yes, load and summarize. If no, start fresh.
  2. Policy mode — (skip if no libraries specified) Ask: prefer libraries with built-in fallback, or custom libraries blocks only (no fallback)?
  3. Excluded blocks (optional) — Are there any blocks (library or built-in) that should never be placed in your models? Collect referenceBlock + reason.
  4. Off-limits parameters (optional) — Are there any block parameters you don't want the agent to modify? Collect referenceBlock + protectedParams[].
  5. Save — Validate and write with library.BlockPolicy.save().

API

policyData = struct();
policyData.policyMode = 'approved_blocks_only';
policyData.fallbackToBuiltins = false;
% Excluded — agent will never place these blocks
policyData.blockedBlocks = struct('referenceBlock', 'LegacyLib/OldCtrl', 'reason', 'Superseded');
% Protected params — agent won't modify these values
policyData.blockRules = struct('referenceBlock', 'Lib/SpeedCtrl', ...
    'protectedParams', {{'Kp', 'Ki', 'Kd'}});

library.BlockPolicy.save(projectRoot, policyData);

For defaults (no restrictions):

policyData = library.BlockPolicy.defaults();
library.BlockPolicy.save(projectRoot, policyData);

Schema

{
  "schemaVersion": 1,
  "policyMode": "approved_blocks_only | prefer_customer_libraries",
  "fallbackToBuiltins": false,
  "blockedBlocks": [
    { "referenceBlock": "LegacyLib/OldController", "reason": "Superseded" }
  ],
  "blockRules": [
    { "referenceBlock": "Lib/SpeedController", "protectedParams": ["Kp", "Ki", "Kd", "SampleTime"] }
  ]
}

Validation Rules

  • policyMode: must be prefer_customer_libraries or approved_blocks_only
  • approved_blocks_only + fallbackToBuiltins: true → rejected (contradictory)
  • protectedParams must be a non-empty array of strings
  • Malformed policy → error (never silently ignored)

Updating an Existing Policy

When .satk/block-policy.json already exists:

  1. Load with library.BlockPolicy.loadRaw(projectRoot, libConfig)
  2. Summarize current state — do NOT re-run the full wizard
  3. Ask what to change, apply to existing struct, save

Common updates: exclude a block, mark params as off-limits, change mode, remove an entry.

Guardrails

  • Never write block-policy.json without user confirmation
  • Always validate before saving (library.BlockPolicy.validate(policyData))
  • If user provides a block name without the full library path, look it up in the KG to resolve the referenceBlock

Copyright 2026 The MathWorks, Inc.


Signals

GitHub stars
1k
Forks
103
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
configuring-block-policy
Source
github.com/matlab/simulink-agentic-toolkit