What is control-metalayer-loop?

A control-system metalayer to drive autonomous code-agent development with explicit primitives, governance, and scalable topology.

What profiles exist and what do they include?

baseline provides a minimal harness; governed adds policy/commands/topology and metrics; autonomous includes recovery, nightly controls, and web/CLI end-to-end capabilities.

How do I start and evolve the system?

Initialize with the desired profile using the control_wizard, audit regularly, and scale by adding primitives and governance while monitoring metrics and entropy.

control-metalayer-loop

Scanned

npx machina-cli add skill broomva/agent-control-metalayer-skill/control-metalayer-loop --openclaw

Files (1)

SKILL.md

3.5 KB

Control Metalayer Loop

Use this skill to initialize or upgrade a repository into a control-loop driven agentic development system.

What To Load

references/control-primitives.md for the control model and minimal control law.
references/rules-and-commands.md for policy/rules and command governance.
references/topology-growth.md for repository topology and scale path.
references/wizard-cli.md for command usage.

Primary Entry Point

Use the Typer wizard:

python3 scripts/control_wizard.py init <repo-path> --profile governed

Profiles:

baseline: minimal harness and command surface.
governed: baseline + policy/commands/topology + control loop + metrics + git hooks.
autonomous: governed + recovery/nightly controls + web and CLI E2E primitives.

Workflow

Baseline current repo workflows and constraints.
Initialize baseline metalayer artifacts.
Add control primitives and governance rules.
Audit and close gaps.
Iterate based on run outcomes and metric drift.

Step 1: Baseline

Identify canonical test/lint/typecheck/build commands.
Identify high-risk actions requiring policy gates.
Identify required observability IDs for agent runs.

Step 2: Initialize Metalayer

Run:

python3 scripts/control_wizard.py init <repo-path> --profile baseline

This creates stable operational interfaces:

AGENTS.md, PLANS.md, METALAYER.md
Makefile.control and scripts/control/*
docs/control/ARCHITECTURE.md and docs/control/OBSERVABILITY.md
CI workflow for control checks

Step 3: Add Control Primitives

Run:

python3 scripts/control_wizard.py init <repo-path> --profile governed

This adds the core control plane:

.control/policy.yaml
.control/commands.yaml
.control/topology.yaml
docs/control/CONTROL_LOOP.md
evals/control-metrics.yaml

For a fully self-sustaining loop:

python3 scripts/control_wizard.py init <repo-path> --profile autonomous

Adds:

scripts/control/install_hooks.sh + .githooks/*
scripts/control/recover.sh
scripts/control/web_e2e.sh
scripts/control/cli_e2e.sh
.github/workflows/web-e2e.yml
.github/workflows/cli-e2e.yml
tests/e2e/web/* + playwright.config.ts
tests/e2e/cli/smoke.sh
.control/state.json
.github/workflows/control-nightly.yml

Step 4: Validate

Run:

python3 scripts/control_wizard.py audit <repo-path>
python3 scripts/control_wizard.py audit <repo-path> --strict

Treat audit failures as blocking until corrected.

Step 5: Operate And Grow

Keep command names stable (smoke, check, test, recover).
Keep E2E command names stable (web-e2e, cli-e2e).
Keep policy and command catalog synchronized with actual behavior.
Track control metrics and adjust setpoints deliberately.
Prune stale rules/scripts/docs to prevent entropy growth.

Adaptation Rules

Do not overwrite existing project conventions without explicit reason.
Prefer wrappers and policy files over ad-hoc command execution.
Make every major behavior observable and auditable.
Keep human escalation rules explicit and easy to trigger.

Source

git clone https://github.com/broomva/agent-control-metalayer-skill/blob/main/.skills/control-metalayer-loop/SKILL.mdView on GitHub

Overview

Init or upgrade a repository into a control-loop driven agentic development system. It provides explicit control primitives, governance, and a scalable topology to keep agents safe while enabling continuous improvement over time.

How This Skill Works

Use the Typer wizard to initialize or upgrade a repo with control-plane artifacts. It creates stable interfaces, policy and command catalogs, topology, and observability hooks, then wires in CI and docs under .control and references, enabling a governed lifecycle for agent development.

When to Use It

When you need explicit control primitives (setpoints, sensors, controller policy, actuators) and a feedback loop for stability and entropy controls
When repository governance is required via policies, commands, and strict command governance
When you want a scalable folder topology that lets agents operate safely and improve over time
When you need multiple profiles (baseline, governed, autonomous) to match risk tolerance and operational maturity
When you must audit, validate, and track control metrics and gaps to prevent drift

Quick Start

Step 1: python3 scripts/control_wizard.py init <repo-path> --profile baseline
Step 2: python3 scripts/control_wizard.py init <repo-path> --profile governed
Step 3: python3 scripts/control_wizard.py audit <repo-path> [--strict]

Best Practices

Keep command names stable (smoke, check, test, recover) across updates
Synchronize policy/command catalogs with actual behavior
Make major behaviors observable and auditable
Prefer wrappers and policy files over ad-hoc command execution
Prune stale rules, scripts, and docs to prevent entropy growth

Example Use Cases

Initialize a repository with the baseline profile to create stable interfaces (AGENTS.md, PLANS.md, METALAYER.md) and Makefile.control
Upgrade to governed profile to add core control plane (.control/policy.yaml, .control/commands.yaml, topology, docs) and governance
Move to autonomous profile to enable recovery, nightly controls, web/CLI E2E primitives and related workflows
Run audits to validate control surfaces and fix failures with python3 scripts/control_wizard.py audit <repo-path> and optional --strict
Operate and grow by tracking metrics, adjusting setpoints, and pruning entropy to keep behavior observable and auditable

Frequently Asked Questions

Add this skill to your agents