gosmith

Create and optimize specialized Claude Code agents. Expertise in agent design, prompt engineering, domain modeling, and best practices. Use PROACTIVELY when designing new agents or improving existing ones.

Shared DevSecOps, secrets management, and container build standards for CI/CD agents (tekton-expert, gha-expert).

Shared policies for Claude Code extensibility. Referenced by agent-expert, skill-expert.

Kubernetes and OpenShift cluster specialist for platform administration, operator development, and security hardening.

Design and optimize token-efficient Claude Skills.

Produces README, API docs, and architecture documentation from code and PLAN.md. Invoke AFTER implementation is stable.

Tekton and Kubernetes-native CI/CD specialist for pipelines, GitOps workflows, and automated delivery.

Manages context across multiple agents and long-running tasks. Use PROACTIVELY when coordinating complex multi-agent workflows or when context needs to be preserved across multiple sessions. MUST BE USED for projects exceeding 10k tokens.

Behavioral guardrails for all agents. Prevents scope creep, overengineering, and silent assumptions.

GitHub Actions specialist for workflows, CI/CD automation, and GitHub-native DevSecOps.

Invoke FIRST for any Go project change requiring design decisions. Produces a PLAN.md blueprint with interfaces, data models, and implementation steps. Other agents then execute against this plan.

Common policies for Go agents. Referenced by go-architect, go-developer, go-tester, go-reviewer.

Implements Go code from PLAN.md blueprints or handles simple tasks independently. Primary agent for writing production code.

Reviews PRs for logic errors, security flaws, and PLAN.md compliance. Invoke BEFORE merging changes from go-developer.

Validates Go implementations against PLAN.md contracts. Invoke AFTER go-developer has written production code.

Shared policies for infrastructure agents (openshift-expert, tekton-expert). Referenced for collaboration patterns and common standards.

Meta-Cognitive Reasoning Framework (MCRF). Referenced by all agents for structured reasoning.

5-element persona framework template for creating Claude Code agents

Fetch Jira tickets as clean markdown using native ADF format.

Run, discover, or review Makefile targets. Use for build automation.

Fetch PR review comments, triage against codebase, apply fixes, generate reply report.

Verify answers by checking sources, contradictions, and confidence level.

Extract readable web content, removing navigation/ads to save tokens.

Run Ginkgo BDD tests with Kubernetes-aligned conventions.

Run Go verification pipeline (build + test + vet + lint).

Go code quality standards, idioms, and YAGNI/KISS development principles.

Generate Go test coverage report with package breakdown.