skills.rest
sakamotomomotaro0809-netizen
sakamotomomotaro0809-netizenCOMMUNITY
Explore all skills by this author →
GitHub

sdd-guardrails

Generate AI agent guardrails documents defining permission boundaries, resource limits, and approval gates.

Updated Mar 23, 2026
Search Tags:#security#permissions#ai agent#audit trail#spec-driven development#human-in-the-loop#guardrails
One-click install
npx skills add https://github.com/sakamotomomotaro0809-netizen/tateyomi --skill sdd-guardrails-sakamotomomotaro0809-netizen
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sdd-guardrails
Source: https://github.com/sakamotomomotaro0809-netizen/tateyomi/tree/main/taisun_agent/.claude/skills/sdd-guardrails
Command: npx skills add https://github.com/sakamotomomotaro0809-netizen/tateyomi --skill sdd-guardrails-sakamotomomotaro0809-netizen

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI coding agents operating without explicit constraints can perform unintended file deletions, run dangerous commands, or access secrets. This Skill produces a structured guardrails.md document that enforces least-privilege permissions, human approval gates, and audit logging for agent-driven development workflows. ## Core Features & Use Cases - Permission Boundaries: Generates file-system, command-execution, and API-call allow/deny tables based on the project's requirements.md and threat-model.md. - Human-in-the-Loop Gates: Defines which operations (production deploys, DB writes, credential access) always require human approval versus which are auto-allowed. - Resource Limits & Audit Trail: Specifies execution time, file size, and API call caps plus JSON-structured logging requirements for every tool invocation. - Use Case: After writing requirements and a threat model for a new spec in .kiro/specs/, run the skill to produce guardrails.md that constrains all subsequent implementation agents. ## Quick Start Ask the agent to generate guardrails for the spec slug google-ad-report using its existing requirements.md and threat-model.md files.

Frequently Asked Questions about sdd-guardrails

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

FAQPage Schema
How do I generate guardrails for an AI coding agent?▼

Run the skill with a spec slug, for example /sdd-guardrails google-ad-report. It reads requirements.md and threat-model.md from .kiro/specs/<slug>/ and writes a guardrails.md covering permissions, limits, approval gates, and audit logging.

What inputs does the guardrails generator require?▼

It requires an existing requirements.md and threat-model.md in the target spec directory. The security requirements and threat mitigations from those files drive the permission boundaries and approval gates in the output.

What is a human-in-the-loop gate for AI agents?▼

A human-in-the-loop gate requires explicit human approval before the agent performs sensitive operations such as production deploys, production database writes, or credential access. Routine actions like editing source files or running tests are auto-approved.

Can I customize which commands the agent is allowed to run?▼

Yes. The generated guardrails.md contains command execution tables listing allowed commands like npm run build and git commit, and blocked commands like git push --force, rm -rf /, and sudo. You can edit these tables after generation.

What happens when the agent hits a permission error?▼

The generated error handling rules specify that permission errors and resource limit overruns cause an immediate stop with mandatory notification, while external API failures trigger up to three retries with optional notification.

Related Skills

Explore More Software Engineering Skills→
gaiastudio-aigaiastudio-ai
2

gaia-threat-model

Automate STRIDE and DREAD threat modeling into a validated threat-model document.

Official
Advanced
CsHengCsHeng
10

config-sync-cli-workflow

Orchestrate multi-target CLI configuration synchronization with phase-driven workflows and backups.

Community
Advanced
COG-GTMCOG-GTM

guardrail-auditor

Monitor Devin Enterprise Guardrails API violations and block non-compliant merges.

Official
Advanced
KingxiaoKingxiao

agentforge-security

Design multi-layer security for autonomous agents with Starlark policies.

Community
Advanced