jlave-dev
- 22 skills
- 0 followers
- 11 hours ago last updated
- ▌ Yagni · jlave-dev bundlePlan or perform ruthless YAGNI repo cleanup. Use when asked to identify or remove stale docs, unused code/deps, duplicate scripts/config/deploy paths, dead flags/routes, speculative abstractions/product surfaces, misleading tests, or implementation/docs drift while preserving active decisions and guardrails.
- ▌ Find Orders · jlave-dev bundleUse when the user asks to find Amazon orders, invoices, return windows, warranty details, reorder links, delivery status, or purchase history. Works from the user's signed-in Amazon account when available and treats order work as read-only unless the user explicitly confirms a separate account-changing action.
- ▌ Shop Amazon · jlave-dev bundleUse when the user asks to shop on Amazon, compare Amazon products, research purchase options, or build a product shortlist. Verifies product-page ASIN/title/price/rating/review data and ranks options by user criteria, favoring review count when ratings are close.
- ▌ Save Note · jlave-dev bundleSave relevant thread context as Markdown files in a local notes directory. Use when the user asks to save a note, capture or remember conversation context, preserve decisions, summarize an active thread for later, or turn explicit, implied, or inferred thread context into a local .md note. For durable planning documents, use save-plan instead.
- ▌ Save Plan · jlave-dev bundleSave durable planning documents as Markdown files in $HOME/plans. Use when the user asks to save a plan, write this down as a plan, put planning work in the plans directory, preserve a migration/design/review/implementation plan, hydrate an agreed plan with detail, or update an existing saved plan.
- ▌ Write Reviews · jlave-dev bundleUse when the user asks to draft, refine, or submit an Amazon product review for a purchased item. Verifies the product/order context when possible, drafts in the user's requested tone, and submits only after explicit user approval.
- ▌ Write Prompt · jlave-dev bundleTurn rough intent into a copy-ready prompt sized to the task. Use when the user asks to write or refine a prompt, invokes $write-prompt, or says promptify; not to execute the underlying task.
- ▌ Create Readme · jlave-dev bundleCreate or update accurate README.md files. Use when asked to draft, rewrite, audit, or refresh a README for a repository, package, app, library, CLI, service, dataset, or docs folder.
- ▌ Sdlc Evidence · jlave-dev bundleUse when an Agent SDLC implementation has a PR or review verdict and the workflow needs current-head checks, local verification, attached proof, simulator evidence, blocker classification, PR body updates, or GitHub issue Agent State refresh before review or merge.
- ▌ Sdlc Reviewer · jlave-dev bundleUse when acting as an independent software reviewer for an SDLC thread workflow, especially when another thread sends a code-change handoff and expects severity-ranked findings plus an explicit review verdict.
- ▌ Sdlc Preflight · jlave-dev bundleUse when an Agent SDLC issue is ready to dispatch or resume and the orchestrator must verify the issue, dependency mode, branch, worktree, base ref, active issue/PR overlap, and hard-stop conditions such as detached HEAD or wrong checkout.
- ▌ Create Agents Md · jlave-dev bundleCreate, update, audit, or split AGENTS.md files for repositories and subdirectories. Use when Codex needs repo-specific agent instructions for commands, conventions, validation, architecture boundaries, monorepo routing, safety, or review workflows.
- ▌ Refactor UI · jlave-dev bundleUse when editing frontend code to remove clutter, leaky copy, redundant UI, duplicate actions, nested cards, or over-explained screens while preserving behavior.
- ▌ Sdlc Merge Queue · jlave-dev bundleUse when Agent SDLC PRs are approved and marked merge-ready, and an orchestrator should merge one or more PRs in dependency order while verifying current-head checks, issue Agent State, review verdicts, and handing conflicts or failed checks back to workers instead of fixing inline.
- ▌ Sdlc Orchestrate · jlave-dev bundleUse when the user wants Agent SDLC to continue an issue or pull request to the next workflow role automatically, including preflight, dispatch, evidence refresh, review, docs, or merge queue routing from GitHub issue/PR state.
- ▌ Sdlc Review Loop · jlave-dev bundleUse when running an Agent SDLC implementer/reviewer loop for a git-backed change, including generating a handoff, reusing the singleton reviewer thread, classifying review verdicts, iterating fixes, and keeping evidence/issue state current.
- ▌ Sdlc Issue Intake · jlave-dev bundleUse when turning a rough feature request, bug report, cleanup idea, or agent task into a scoped GitHub Issue for Agent SDLC workflows, including active-work overlap checks, dependency classification, labels, branch naming, CI tiering, and a worker dispatch prompt saved into the issue.
- ▌ Reason From Evidence · jlave-dev bundleInvestigate evidence with explicit hypotheses and justified conclusions. Use when debugging systems, analyzing incidents or reports, reducing possible explanations, finding missing evidence, or separating proof, consistency, probability, and speculation.
- ▌ Sdlc Dispatch Issue · jlave-dev bundleUse when an orchestrator should pick or receive an Agent SDLC GitHub Issue whose preflight has passed, create the implementation branch/worktree, spawn the worker thread from the issue's saved Worker Dispatch prompt, or record worker state back on the issue without doing worker fixes inline.
- ▌ Map Project Decisions · jlave-dev bundleMap a greenfield software project's components and separate human-owned product or architecture decisions from safe, reversible, or testable choices the agent should make. Use when starting a new app, service, CLI, integration, or major subsystem.
- ▌ Audit Rendered UI · jlave-dev bundleUse when reviewing screenshots or rendered frontend UI for leaky copy, redundancy, visual clutter, responsive fit, broken assets, or handoff polish.
- ▌ Reference Ruler UI · jlave-dev bundleUse when matching a frontend implementation to a concrete visual reference: image-to-code UI work, mockup or screenshot replication, screenshot-driven polish, pixel-fidelity repair loops, or generated mockups that need faithful implementation.