Detect session degradation and spawn Haiku subagents for code review.
npx skills add https://github.com/Overlord-Z/ClaudeShack --skill guardian-overlord-z
Please help me install this Agent Skill. Skill: guardian Source: https://github.com/Overlord-Z/ClaudeShack/tree/main/skills/guardian Command: npx skills add https://github.com/Overlord-Z/ClaudeShack --skill guardian-overlord-z
Guardian automates session health monitoring and provides automatic code review suggestions when degradation is detected, while keeping changes read-only and validated via Oracle.
To use Guardian, run: python guardian.py review --file <path> --focus <topic>, or check triggers with python guardian.py check
High-intent search queries and answers about installing and using this skill.
Session health monitoring detects code degradation through volume, error patterns, and file churn. Guardian triggers automated code review when thresholds are exceeded, surfacing issues before they compound.
Yes. Guardian spawns Haiku subagents to inspect code with minimal-context analysis, reducing token usage and latency while maintaining review depth for authentication, security, and complex task contexts.
Oracle validation gates all suggestions. Guardian learns from your feedback to adjust sensitivity, ensuring only user-approved interventions surface and read-only operations remain safe.
Code review triggers on repeated errors, file churn, correction patterns, and code volume thresholds. Guardian monitors session health continuously and spawns targeted subagents when degradation is detected.
Guardian integrates with Oracle for validation, Evaluator for metrics, and Summoner for orchestration, fitting into workflows that require governance, minimal context passing, and continuous quality gates.
Guardian performs read-only analysis and requires user approval before changes. It works best in software development workflows with authentication and security concerns where minimal-context analysis across multiple tasks is needed.
wientjesDiagnose NanoClaw container agent authentication, mount, and log failures.
writerGenerate plain-language narratives explaining population health trends from clinical, utilization, and cost data.
Overlord-ZCoordinates multi-agent orchestration for task decomposition, quality gates, and production-ready code.
jtsang4Execute detailed implementation plans in batches with review checkpoints.
