Columbus workflow with a fast indexed memory system, reusable skills, agents, and hooks.
On-demand codebase explorer and context-retrieval task agent. Uses Columbus CLI search, graphs, memory, symbol, and file projections to return one enriched, LLM-ready report. Use when the user, another agent, or an orchestrated workflow needs grounded codebase context, implementation locations, dependency shape, or relevant recorded decisions.
Implements one scoped piece of Columbus-planned work with minimal, verified code changes. Use when a plan step has clear acceptance criteria and needs code edits, command execution, and tests.
Designs and runs verification for scoped Columbus-planned work. Use when adding tests, checking acceptance criteria, strengthening regression coverage, validating bug fixes, or deciding whether planned work can be called done.
Reviews scoped changes for correctness, maintainability, regressions, and fit with the planned scope. Use after implementation or before calling Columbus-planned work done.
Reviews scoped changes for vulnerabilities, dependency/CVE exposure, secrets, auth, authorization, input validation, injection, and data-handling risk. Use for security gates before shipping planned work or releases.
Teaches how Columbus CLI works as a local semantic code-context and durable memory system, with references for commands, retrieval, memory (adr, plan, documentation), setup, and agent usage. Use when the user asks how Columbus works, wants to learn Columbus, needs Columbus command examples, asks about search, indexing, memory, ADRs, plans, documentation memories, navigator usage, or says "teach me Columbus".
Create git commits using Conventional Commits style, and never add co-authorship or tooling attribution to the message. Use when committing changes, writing or amending commit messages, splitting work into commits, or when the user says "commit", "/commit", or asks to stage and commit work.
Survey any codebase as a senior advisor and produce prioritized, self-contained implementation plans recorded as Columbus plan memories for OTHER models/agents to execute. Strictly read-only on source code — never implements, fixes, or refactors anything itself. Use when asked to audit a codebase, find improvement opportunities (bugs, security, performance, test coverage, tech debt, migrations, DX), suggest features or where to take the project next (roadmap, product direction), or generate handoff plans for another agent to implement.
Interview session that interrogates your plan against the project's domain model and recorded decisions, sharpens terminology, and records or updates Columbus memory (ADRs and documentation) inline as the plan crystallises. Use when the user wants to stress-test or interview a plan against the codebase's language and documented decisions, or says "interview me", "grill my plan", or "challenge this design".
Master professional UI design and implementation — design tokens and theming with Tailwind CSS v4, shadcn/ui components, UX principles, micro-interactions, animation with Motion (formerly Framer Motion), color science (OKLCH, contrast, dark mode), and typography. Use when building or improving user interfaces, when asked to make a UI "look professional", "polished", or "better", when styling with Tailwind or shadcn, designing color palettes or dark mode, choosing fonts, adding animations, transitions, or motion design, building or auditing a design system, or reviewing UX and accessibility.
Executes bash commands
Hook triggers when Bash tool is used
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
This repository hosts the Columbus Workflow Claude Code plugin, distributed through the
orafaelfragoso marketplace. The marketplace manifest lives at .claude-plugin/marketplace.json,
the plugin manifest at .claude-plugin/plugin.json, and all plugin assets under
plugins/columbus-workflow/.
.
├── .claude-plugin/
│ ├── marketplace.json # Marketplace catalog (orafaelfragoso)
│ └── plugin.json # Plugin manifest
└── plugins/columbus-workflow/ # Plugin assets
├── agents/
├── config/claude/settings.local.example.json
├── output-styles/caveman.md
├── scripts/
└── skills/
Install from the hosted marketplace:
claude plugin marketplace add orafaelfragoso/agentic-workflow
claude plugin install columbus-workflow@orafaelfragoso
When working from a local clone, add the marketplace from the repository root instead:
claude plugin marketplace add ./
claude plugin install columbus-workflow@orafaelfragoso
When developing locally, validate the plugin from the repository root:
claude plugin validate .
Repo consistency checks live in scripts/check-consistency.sh; run them before committing. CI runs them automatically on every push and pull request.
Claude Code loads assets through root manifest paths that point into plugins/columbus-workflow/.
Add new agents to the agents list in .claude-plugin/plugin.json.
This plugin organizes delivery around Columbus plan memories:
ship executes Columbus plan memories through direct, sequential, or parallel delivery flows. Live execution state lives in the session; Columbus holds the durable record.navigator is the on-demand codebase explorer. It can be invoked directly or by orchestration, uses Columbus to locate relevant code, and returns one cited report.adr, shipped behavior as documentation, and executed plans are re-kinded or removed.ship keeps its top-level instructions short and loads one of three references based on the work shape:
plugins/columbus-workflow/skills/ship/references/direct.mdplugins/columbus-workflow/skills/ship/references/sequential.mdplugins/columbus-workflow/skills/ship/references/parallel.mdplugins/columbus-workflow/scripts/prompt.sh is loaded on SessionStart for startup, clear, and compact events. It intentionally emits only compact working rules: Columbus retrieval, plan-driven delivery with ship, delegation, change/verification discipline, git safety, and current external-doc/security checks.
Keep this prompt stable and compact because it is injected into every session.
The bundled agents are flat specialists. ship coordinates the flow; agents do not coordinate each other by default.
The agents pin Claude model aliases (haiku/sonnet), reference Claude MCP tool ids (mcp__context7__*), and use agent-teams tools (SendMessage, EnterWorktree, ExitWorktree) that require CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 — set in the local config template.
navigator: on-demand codebase explorer using Columbus.delivery-engineer: implements one scoped piece of planned work.test-engineer: designs and runs verification.quality-reviewer: reviews correctness, maintainability, regression risk, and fit with the planned scope.security-analyst: reviews vulnerabilities, dependency/CVE exposure, secrets, auth, and data-handling risk.architecture-reviewer: reviews patterns, boundaries, coupling, and abstractions.release-coordinator: checks branch, merge, PR, verification, and plan-closeout readiness.npx claudepluginhub orafaelfragoso/agentic-workflow --plugin columbus-workflowHarness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Claude harness - A harness for solo developers (Vibecoders) to handle full-cycle contract development.
Superpowers Plus core skills library for Claude Code: planning, execution routing, TDD, debugging, and collaboration workflows
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
AI-powered development tools for code review, research, design, and workflow automation.
Unity Development Toolkit - Expert agents for scripting/refactoring/optimization, script templates, and Agent Skills for Unity C# development