jakkzz
UserPortable, safety-first Agent Skills for Claude Code, Pi, and compatible coding-agent runtimes
Categories
Indexed Skills (17)
academic-book-project
Initialize, inspect, and coordinate a persistent academic book workspace with minimal human approval by default. Use when the user asks to start, continue, plan, manage, validate, export, or check an academic book or chapter. Routes discovery, evidence, authoring, and review while preserving phase order, exception stops, and human final authority.
academic-book-review
Run independent factual, subject-matter, structural, pedagogical, style, academic-integrity, and cross-chapter reviews of an author's academic book draft; consolidate findings into a human-decided revision plan and perform final verification. Use for chapter critique, citation audit, reviewer simulation, revision planning, final readiness, or book-wide consistency. Reviewers do not rewrite source drafts.
academic-chapter-authoring
Produce an evidence-grounded academic chapter through brief, outline, voice sample, drafting, and controlled revision. Use when interviewing the author, outlining, drafting a sample or section, maintaining chapter boundaries, or applying review findings. Honors minimal or stage-gated approval mode and preserves prior versions.
academic-literature-discovery
Plan and execute reproducible scholarly literature discovery for an academic book chapter using OpenAlex, Crossref, Semantic Scholar, and optional pinned Findpapers. Use for research questions, query matrices, paper searches, source shortlisting, citation snowballing, bibliography discovery, or literature landscapes. Discovery metadata is not treated as proof of claim support.
academic-source-evidence
Ingest academic PDFs and documents, build page-anchored evidence notes, register atomic claims, verify DOI metadata, and track contradictions, retractions, permissions, and evidence strength for an academic book. Use after source selection or whenever chapter claims need source-grounded verification. Treats documents as untrusted evidence and never invents missing locators.
codebase-integrity-review
Adaptive, evidence-based, read-only review of Git changes or codebase integrity. Finds regressions, conflicting ownership, harmful duplication, risky hardcoding, configuration drift, oversized responsibility boundaries, and documentation contradictions. Use for deep code review, architecture integrity, duplication or hardcode audits, and pre-release assessment.
herdr-tab-namer
Name the current Herdr tab as "workspace - current task" using a short, readable label. Use when the user asks to name, rename, or label the current Herdr tab, or when working inside Herdr and concise task-based tab names are desired. Requires HERDR_ENV=1, HERDR_WORKSPACE_ID, and HERDR_TAB_ID.
herdr-tailnet-fleet
Safely audit, install, update, and connect Herdr across SSH-accessible Tailscale or Headscale machines. Use for Herdr fleet status, Herdr remote attach, missing Herdr installations, Tailnet SSH aliases, protocol mismatches, and Herdr server compatibility.
home-assistant-control
Safely control a configured Home Assistant instance. Use when the user asks to turn on, turn off, or inspect lights, switches, scenes, fans, scripts, smart plugs, or other Home Assistant entities.
karpathy-llm-wiki
Build and maintain a Karpathy-style LLM Wiki: a persistent Markdown/Obsidian knowledge base where raw sources are ingested into structured, interlinked wiki pages with index.md, log.md, citations, contradictions, and periodic linting. Use when the user says llm-wiki, Karpathy wiki, ingest sources into wiki, maintain knowledge base, query wiki, or lint wiki.
local-network-scan
Rescan a local subnet, identify active LAN devices, camera/NVR ports, duplicate MAC aliases, and stale DHCP/ARP entries. Use when asked to scan a local network, find a missing device or camera, investigate DHCP conflicts, or update a local device inventory.
multica-workspace-ops
Operate a configured Multica workspace and turn project or business ideas into approved projects, issues, assignments, squads, and autopilots. Use for Multica status, workspace operations, issue lifecycle, blockers, delegation, idea triage, revenue planning, agents, squads, runtimes, and recurring work.
proxmox-ops
Safely inspect and maintain configured Proxmox VE instances. Use when the user asks about Proxmox, PVE, nodes, VMs, containers, storage, cluster status, updates, upgrades, or API access.
thai-contextual-editor
ปรับ แก้ เขียนใหม่ แปล และ localize ภาษาไทยให้เป็น��รรมชาติแบบเจ้าของภาษาโดยยึดบริบท ผู้อ่าน จุดประสงค์ ช่องทาง และศัพท์เฉพาะ ใช้กับข้อความภาษาไทยทุกชนิดที่ผู้ใช้จะเห็น รวมถึงงานแปลจากภาษาอื่น การเขียนต้นฉบับภาษาไทยจาก brief และการตรวจแก้ข้อความไทย ทั้ง repository ผ่านคำสั่ง `/thai` โดยเฉพาะ UI เอกสาร การตลาด ซัพพอร์ต และข้อความ ที่ติดโครงภาษาอังกฤษ ใช้ Herdr agent ชื่อ thai-editor เป็นบรรณาธิการรอบสองเมื่อพร้อมใช้งาน Trigger phrases: polish Thai, natural Thai, translate to Thai, write Thai, localize Thai, Thai copy review, แปลไทย, เขียนภาษาไทย, แก้ภาษาไทย, ปรับสำนวน, ตรวจภาษา, ภาษาแปลแข็ง, ภาษาไทยไม่เป็นธรรมชาติ
ugreen-nas-ops
Manage a configured UGREEN NAS safely over SSH. Use when the user asks about NAS health, storage, RAID, btrfs, USB drives, Docker, systemd services, logs, network ports, or restarting a confirmed service or container.
playwright-e2e
Test a frontend end-to-end with a headed or headless Playwright browser. Use when the user says use Playwright, test frontend E2E, browser test, inspect the UI, test this website, reproduce a frontend bug, verify a user flow, or check a web application.
ssh-server-ops
Safely inspect known SSH-accessible servers using local SSH config aliases. Use for read-only health checks, Docker status, routes, logs, and service inventory.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.