yansfil
- 10 skills
- 0 followers
- 15 hours ago last updated
- ▌ Ship · yansfil bundleDeliver a completed `implement` receipt locally or through GitHub PR delivery. Use when the user invokes "$ship" or explicitly asks to create a local implementation commit, open or update a PR, push a completed PRD implementation, watch CI, or merge an approved green PR through the recorded delivery workflow. Do not use before a complete `implement` receipt exists.
- ▌ Quick · yansfilSession-context quick pipeline: compact contract, implement, verified receipt. Use when the user invokes "$quick", asks to implement something small from the current conversation without the PRD pipeline, or wants the shortest path with required checks and an independent whole-contract review. Do not use when a PRD exists or the user asks for the PRD pipeline.
- ▌ Please · yansfilAll-in-one PRD pipeline runner. Use when the user invokes "$please", asks to take the current conversation and carry it through to a finished implementation in one shot, wants the gen-prd then implement then ship chain run automatically without approval round-trips, or says things like "그냥 끝까지 해줘", "대화한 대로 구현까지 해줘", "one shot implement this".
- ▌ Gen Prd · yansfil bundleProject-local PRD writer. Use when the user invokes "$gen-prd", asks for a PRD, product requirements document, implementation-ready requirements, or wants to turn intake/clarify output into a human-reviewable contract of about 100 lines: a goal, non-goals, a Decisions table, a complete Behaviors table with decision references, a short technical structure, and risks.
- ▌ Remember · yansfilLand a lesson as an enforcement asset so the same mistake cannot repeat. Use when the user invokes "$remember", says "remember this", "learn from this", "don't do that again", "배워둬", "기억해둬", "같은 실수 하지마", when a finalize receipt suggests lessons from recorded deviations, or when a painful debugging/rework session just ended and its cause is nameable.
- ▌ Challenge · yansfilRed-team a conclusion that is already on the table: take the dissenting opinion seriously, investigate it on its own terms, then rule on it. Use when the user invokes "$challenge", types the "!rv" trigger, says "반박해봐", "이거 아니라는데", "다시 검증해봐", "red team this", "challenge this", "steelman the objection", or when a decision is about to be acted on and someone has pushed back on it. Do not use to review an unreviewed diff - that is what the gates are for.
- ▌ Implement · yansfil bundleProject-local approved-PRD implementation executor and Observer entrypoint. Use when the user invokes "$implement", explicitly asks to execute an approved PRD through the receipt-backed workflow, or wants the complete PRD implemented, independently reviewed, and finalized with current evidence. Do not use for ordinary implementation requests that have no approved PRD.
- ▌ Sasu Setup · yansfilProject-local PRD pipeline configuration. Use when the user invokes "$sasu-setup", asks to enable or change PR delivery mode, configure worktree/secrets sync for implement, configure the agents/ namespace gitignore policy, inspect the current PRD pipeline settings, or diagnose why implement/ship delivery is not working.
- ▌ Interview Me · yansfil bundleProject-local pre-PRD requirements interview. Use when the user invokes "$interview-me", asks to interview an idea, clarify or pressure-test requirements, reduce ambiguity, prepare a PRD source, or wants UX, behavior, scope, technical, verification, risk, operation, and documented-domain decisions captured before writing a PRD. Run low-latency main-agent-led Q&A with raw capture, decision tracking, targeted UX scenario coverage, periodic checkpoint backfill, and a final PRD-ready qa-log.
- ▌ Benchmark Implement · yansfil bundleRun an approved PRD through the existing receipt-backed implement workflow, analyze its runtime session with a fresh independent evaluator, emit a deterministic process report, and compare it with an earlier run. Use when the user invokes "$benchmark-implement", asks to benchmark the implement skill or harness, rerun a fixed PRD after harness changes, analyze implementation-session efficiency, or compare baseline and candidate harness behavior.