NeuroSkill-com/skills

NeuroSkill™ and NeuroLoop™ markdown files

Compatible avecClaude CodeCodex CLI~Cursor
npx add-skill NeuroSkill-com/skills

name: neuroskill-index description: Skill index and overview — lists all available NeuroSkill skills, tools, and explains how skill loading works. index: true

NeuroSkill Skill Index

NeuroSkill is a biometric AI companion powered by a real-time EEG device. It reads brainwaves and physiology continuously, and uses that data to inform every response.

NeuroSkill exposes the EEG analysis API through a local WebSocket server and HTTP tunnel. The npx neuroskill <command> CLI is the fastest way to query it from a terminal, shell script, or any automation pipeline.

Skills are loaded contextually — the LLM loads the relevant skill file into its context when the user's message matches the skill's domain.


EEG Data & API Skills

SkillLoaded whenDescription
neuroskill-transporttransport/connection questionsWebSocket & HTTP transport, port discovery, Quick Start, output modes (--json / --full), and global CLI flags.
neuroskill-statusstatus/device questionsstatus command — full system snapshot: device state (4 device types), signal quality, EEG scores, band powers, ratios, hooks summary, embeddings, labels, sleep summary, and recording history.
neuroskill-sessionssession/history questionssession and sessions commands — per-session metric breakdowns with first/second-half trends, session listing, and Unix timestamp helpers.
neuroskill-searchcomparison/trend questionssearch and compare commands — ANN search for neurally similar EEG moments across all history, and A/B session comparison with metric deltas and UMAP enqueuing.
neuroskill-sleepsleep/fatigue contextsleep and umap commands — EEG-based sleep stage classification (Wake/N1/N2/N3/REM) with efficiency and bout analysis, and 3D UMAP projection for spatial session comparison.
neuroskill-labelslabel/search contextlabel, search-labels, and interactive commands — creating EEG text annotations, semantic vector search over labels, and cross-modal 5-layer graph search combining text, EEG similarity, and screenshot discovery with 3D visualization.
neuroskill-streamingstreaming/calibration/TTS contextsay, listen, notify, calibrate, calibrations, timer, and raw commands — on-device TTS speech, real-time WebSocket event streaming, OS notifications, calibration profile management, focus timer, and raw JSON passthrough.
neuroskill-hookshooks/automation contexthooks command — Proactive Hooks CRUD, real-time EEG pattern matching with scenario gating (cognitive/emotional/physical), threshold suggestions, WebSocket broadcast triggers, and audit logging.
neuroskill-dndDND/focus-mode contextdnd command — EEG-driven Do Not Disturb automation, rolling focus score average, OS-level DND state, and force-override on/off.
neuroskill-calendarcalendar/schedule/meetings/events questionscalendar command — OS calendar events via EventKit (macOS), iCal files (Linux/Windows); fetch by range, check access status, request permission.
neuroskill-screenshotsscreenshot/OCR/vision search contextsearch-images, screenshots-around, screenshots-for-eeg, and eeg-for-screenshots commands — search screenshots by OCR text (semantic or substring), by visual similarity (CLIP --by-image), find screenshots near a timestamp or EEG session, and cross-modal EEG↔screenshot bridging (find brain state for screen content, or screen content for brain state).
neuroskill-data-referencemetric field questionsAll metric fields — band powers, EEG ratios and indices, core scores, complexity measures, PPG/HRV, motion and artifact markers, sleep stage codes, neurological indices, and consciousness metrics.
neuroskill-recipesscripting/automation questionsUse-case recipes and scripting patterns — focus monitoring, stress tracking, sleep analysis, cognitive load queries, meditation tracking, cross-modal graph search, A/B comparison, time-range queries, and automation with cron/Python/Node.js/HTTP.

LLM & AI Skills

SkillLoaded whenDescription
neuroskill-llmLLM/chat/model questionsBuilt-in on-device LLM inference server — model catalog management (add/download/select/delete GGUF models), vision support (mmproj), streaming WebSocket and HTTP chat, automatic tool calling, GenParams tuning, persistent chat history, and OpenAI-compatible API.

Protocol & Intervention Skills

SkillLoaded whenDescription
neuroskill-evidenceany intervention, protocol, habit tracking, or "what works for me?" questionImplicit evidence collection and personal effectiveness engine — standardised px: label schema, automatic before/after measurement, outcome scoring, personal protocol ranking, life-event tracking, evidence-driven selection rules, and privacy safeguards. All other skills that deliver interventions MUST follow this skill's rules.
neuroskill-protocolsany protocol/exercise/routine intent detectedProtocol framework hub — personalisation engine, API integration guide, modality router, matching guidance, and index of 11 domain sub-skills. Always loaded alongside domain sub-skills.
neuroskill-protocols-focusfocus, attention, cognitive load, energy, alertness, meditation, flow, creativityFocus, attention, cognition, consciousness, deep meditation, and energy/alertness protocols.
neuroskill-protocols-stressstress, anxiety, relaxation, HRV, calm, overwhelm, breathing exercisesStress, relaxation, autonomic regulation, HRV, hemispheric balance, and deep relaxation protocols.
neuroskill-protocols-emotionsemotions, mood, anger, grief, sadness, shame, fear, loneliness, joyEmotional regulation, mood, and extended emotional processing protocols (12 specific emotions).
neuroskill-protocols-sleepsleep, insomnia, nap, rest, recovery, NSDR, tiredSleep, circadian, recovery, NSDR/yoga nidra, and power nap protocols.
neuroskill-protocols-bodybody, tension, neck, headache, eye strain, posture, pain, groundingBody, somatic, neck/cervical, eye exercise, headache/migraine, and motor protocols.
neuroskill-protocols-routinesmorning routine, gym, workout, exercise, hydration, breakMorning routines, workout/gym, hydration, and movement break protocols.
neuroskill-protocols-nutritionfood, diet, nutrition, caffeine, fasting, meal, cravings, alcoholDietary, nutrition, caffeine timing, fasting, eating, and gut-brain protocols.
neuroskill-protocols-musicmusic, playlist, listening, binaural beats, singingMusic protocols — genre/BPM/artist suggestions for mood, focus, stress, sleep, and emotional release.
neuroskill-protocols-digitalsocial media, phone, screen time, scrolling, digital detox, FOMOSocial media, digital addiction, screen time, and attention restoration protocols.
neuroskill-protocols-breathfreenon-breathing, breath-free, alternative, tactile, fidget, cold water30+ non-breathing alternatives — cognitive, tactile, oculomotor, micro-movement, auditory, and passive physiological.
neuroskill-protocols-lifeparent, elderly, teen, student, ADHD, autism, commute, shift work, relationship, disability, culturalContext-specific protocols for 11 life situations — parenting, aging, teens, neurodivergent, commuters, manual workers, healthcare, relational, accessibility, cultural, and situational.

Tools Available to the LLM

The built-in LLM chat has access to these tools (enabled in Settings > LLM > Tools):

ToolPurpose
dateGet current date/time metadata (Unix timestamps, timezone, local/UTC).
locationApproximate public-IP geolocation (country/region/city/timezone).
bashExecute shell commands and return stdout/stderr.
read_fileRead file contents from disk.
write_fileWrite / create files.
edit_fileSurgical find-and-replace edits in files.
search_outputNavigate large tool outputs (paginated regex/head/tail).
web_searchSearch the web and return results.
web_fetchFetch a URL and return its content.
skillQuery the NeuroSkill EEG API (status, sessions, labels, search, hooks, DND, calibrations, TTS, etc.).

How Skill Loading Works

Skills are discovered automatically from:

  1. ~/.skill/skills/ — user-global skills
  2. <project>/.skill/skills/ — project-local skills
  3. The bundled skills/ directory (this repository)

Each skill has a SKILL.md with YAML frontmatter (name, description). On LLM server startup, all skills are discovered and their names/descriptions are injected into the system prompt as an <available_skills> XML block. When the user's task matches a skill's description, the LLM uses the read_file tool to load the full instructions from the SKILL.md file.

Skills associés