skills.rest
hendrax5
hendrax5COMMUNITY
Explore all skills by this author →
GitHub

control

Manage automation levels, pause/resume behavior, and trust scores for bkit-based agents.

Updated Mar 20, 2026
Search Tags:#automation#runtime#guardrails#control#trust-score#audit-log
One-click install
npx skills add https://github.com/hendrax5/ironman --skill control-hendrax5
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: control
Source: https://github.com/hendrax5/ironman/tree/main/skills/control
Command: npx skills add https://github.com/hendrax5/ironman --skill control-hendrax5

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Provides centralized control over an agent's automation level, pause/resume behavior, and trust scoring to prevent uncontrolled or unsafe autonomous actions and to make oversight transparent.

Core Features & Use Cases

  • Status Dashboard: Read and display current automation level, trust score breakdown, active guardrails, paused state, and feature status from runtime state files.
  • Level Management: Validate and set automation levels (L0-L4) with escalation confirmation rules and audit logging for changes.
  • Pause / Resume: Immediately pause automation to L0 and restore previous level safely, preserving previousLevel and audit trails.
  • Trust Reporting: Calculate and present trust score components (PDCA completion, match rate, error recovery, session stability, override frequency) and provide level recommendations.
  • Use Case: An operator reviewing system safety can pause automation during a risky deployment, inspect trust score factors, and then safely resume to the approved level.

Quick Start

Use the control skill to set automation level to L2 and view the current trust score.

Frequently Asked Questions about control

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I pause autonomous agent automation during a risky deployment?▼

You can immediately pause autonomous agent automation to level L0. The system safely preserves your previousLevel state and audit trail entries, allowing you to resume operations once the risk passes.

How does trust score calculation work for automated agent guardrails?▼

Trust score calculation evaluates PDCA completion, match rate, error recovery, session stability, and override frequency. These components generate a breakdown that provides automation level recommendations for safe runtime orchestration.

How do I set automation levels with escalation confirmation in CI/CD workflows?▼

Set automation levels (L0-L4) in CI/CD workflows by validating the requested change and enforcing escalation confirmation rules. The system records audit log entries for every level modification to maintain operational oversight.

Can I view active guardrails and runtime feature states for agent orchestration?▼

Yes, you can view active guardrails and runtime feature states using a status dashboard. It reads control-state and trust-score JSON files to display current automation levels, paused states, and feature status.

What are the limitations of overriding escalation policies in runtime automation control?▼

Overriding escalation policies in runtime automation control is limited by enforced confirmation rules on level increases. High override frequency directly lowers your trust score, which impacts system recommendations for autonomous actions.

Related Skills

Explore More Software Engineering Skills→
solitasrohsolitasroh
2

control

Manage bkit automation levels, trust scores, and pause/resume operations.

Community
Advanced
aberrantCodeaberrantCode
1

existing-repo

Analyze existing repositories to audit tech stack, testing, CI/CD, and documentation patterns.

Community
Intermediate
raphaelchristiraphaelchristi
43

harness:certify

Run repeated evaluations and compute score stability statistics for evolved LLM agents.

Community
Intermediate
gaoqiongxiegaoqiongxie
1

agent-governance

Enforce policies, trust scoring, audit logging, and threat detection for AI Agent workflows.

Community
Intermediate