shuymn
@shuymn
Public Skills
adversarial-verify
by shuymn
"Adversarial verification of a completed task — tries to break the implementation through edge cases, error paths, security probes, and concurrency attacks. Required for Critical/Sensitive tiers (conditionally for Standard with impl files). Use after execute-plan dod-recheck PASS."
setup-ralph
by shuymn
Prepares .ralph/ runtime state from an approved and reviewed plan bundle. Syncs plan tasks into prd.json and updates prompt.run.md with project-specific context and quality gates. Use after plan approval, decompose-plan review PASS, and ralph init.
execute-plan
by shuymn
"Executes one user-selected task from an approved plan bundle using TDD, or rechecks DoD independently (mode=implement dod-recheck). Use when the user specifies a task ID to implement, e.g. \"execute Task 3\" or \"implement task 5\", after decompose-plan review PASS. Also use with dod-recheck mode when the user says \"recheck task DoD\", \"verify task completion\", or \"dod-recheck\"."
decompose-plan
by shuymn
"Decomposes an approved design document into a TDD-based implementation plan or reviews an existing plan (mode=create review). Create mode: task breakdown with traceability. Review mode: independent sub-agent verification absorbing the former analyze-plan audit. Use after design-doc approval when the user needs to break down a design into executable tasks, create an implementation plan, or generate a task list from a design."
commit
by shuymn
Creates meaningful git commits by analyzing changes and committing in logical units. Use when the user wants to commit changes to git, requests commit creation, or asks to save changes to version control. Supports --english and --japanese for commit language selection and --branch to create a new branch before committing.
design-doc
by shuymn
"Creates or reviews a design document (mode=create review). Create mode: iterative dialogue producing design docs and ADRs. Review mode: independent sub-agent verification of an approved design doc. Use this skill whenever a feature, API, architecture, or data model change would benefit from written design before coding — even if the user doesn't explicitly say \"design doc\". When in doubt, invoke this skill first."
audit-agents-md
by shuymn
Audits and refines a CLAUDE.md or AGENTS.md file for instruction density, staleness, and effectiveness. Use when reviewing or improving an agent instruction file, after significant project changes (skills, architecture, or tooling), when agent behavior suggests instructions are ignored or misinterpreted, when the file grows beyond ~30 instruction lines, or when the user says "review my AGENTS.md" or "audit agent instructions".
create-pr
by shuymn
Reviews committed changes and creates a pull request on GitHub. Use when the user wants to create a PR, push changes for review, or open a pull request. Requires a GitHub repository. Supports --japanese flag for Japanese descriptions, --base flag to specify target branch, and --update flag to update an existing PR.
triage-review
by shuymn
"Processes AI reviewer feedback and applies only verified fixes. Works in two modes: (1) fetches comments from a PR URL or current branch, (2) processes feedback pasted directly into the conversation. Trigger when the user wants to bulk-process or apply AI review suggestions — from a GitHub PR or pasted text. Do NOT trigger for single questions about what a bot said, or general code review discussion."
handover
by shuymn
Creates or refreshes a HANDOVER.md that captures current session state for the next assistant. Use when wrapping up a session, context is getting full, switching operators, or when the user asks for a handoff summary including decisions, pitfalls, lessons learned, next steps, and important files. Also trigger when the user says "save progress", "session summary", or "prepare for next session".