oh-my-kiro-cli

Workflow harness for Kiro CLI — skills, agents, and hooks inspired by oh-my-claudecode.

What is this?

A curated collection of Skills, Agents, and Hooks for Kiro CLI.
No extra CLI, no TypeScript runtime — just markdown, JSON, and shell scripts that plug directly into Kiro's native extension points.

oh-my-kiro-cli/
├── skills/     # Skill markdown files (skill:// resources)
├── agents/     # Agent JSON configurations
└── hooks/      # Hook shell scripts

Install

Requirements: Node.js 22+ (LTS or current). Earlier Node versions are no longer security-supported upstream.

| Method | Command | Reviewability | |---|---|---| | npm (recommended) | npm i -g oh-my-kiro-cli@<version> then oh-my-kiro-cli setup | medium | | git clone | git clone ... && less install.sh && ./install.sh | high |

npm (recommended):

npm i -g [email protected]
oh-my-kiro-cli setup

Always pin to a specific version (@0.1.4). Running oh-my-kiro-cli setup is a separate explicit step — nothing is written to ~/.kiro until you run it.

Setup options:

• --with-hooks — also install hook scripts (opt-in)
• --force — overwrite user-owned items in ~/.kiro after creating a .bak.<timestamp> backup
• --link — symlink instead of copy (git-clone installs only)

By default, same-name skills or agents that lack the .oh-my-kiro-managed marker are treated as user-owned and skipped. Pass --force once when upgrading from a pre-0.1.4 install to take over the marker.

git clone (highest reviewability):

git clone https://github.com/hojin-choi/oh-my-kiro-cli ~/.oh-my-kiro-cli
cd ~/.oh-my-kiro-cli
git checkout v0.1.4
less install.sh   # review first
./install.sh

Verifying the npm artifact (optional but recommended for supply-chain auditing):

npm view [email protected] dist.attestations
# Expect: provenance.predicateType = https://slsa.dev/provenance/v1
# Publisher: GitHub Actions (Trusted Publisher), not a personal npm token.

How it works

| Kiro concept | This repo | |---|---| | ~/.kiro/skills/ | skills/*/SKILL.md | | ~/.kiro/agents/ | agents/*.json | | Hook scripts | hooks/*.sh |

Skills are loaded on-demand when Kiro's agent determines they're relevant based on the skill description.
Agents can be switched with /agent inside a Kiro CLI session.

Skills

| Skill | Description | |---|---| | deep-interview | Socratic clarification loop before execution. Use when requirements are vague or ambiguous. | | autopilot | Full autonomous execution from idea to working code across planning, implementation, QA, and validation. | | ralph | PRD-driven persistence loop that keeps working until all acceptance criteria are verified. | | ralplan | Consensus planning with Planner/Architect/Critic loop before any code is written. | | code-review | Parallel two-lane code review (code quality + architecture) with deterministic verdict. | | deepsearch | Multi-strategy codebase search across symbols, usages, and structural patterns. |

Agents

| Agent | Description | |---|---| | architect | Strategic architecture & debugging advisor. Read-only analysis with file:line evidence. | | code-reviewer | Severity-rated code review with SOLID/security checks. Read-only. | | critic | Final quality gate with multi-perspective review and pre-commitment predictions. Read-only. | | debugger | Root-cause analysis with 3-failure circuit breaker. | | executor | Minimal-diff implementation specialist. | | planner | Structured planning consultant. Creates plans only, never implements. | | security-reviewer | OWASP Top 10 analysis and secrets detection. Read-only. | | qa-tester | Behavior verification with evidence capture and clean teardown. |

Contributing

PRs welcome. See CONTRIBUTING.md.

License

MIT