mrdulasolutions

Teaches how a live webpage was built from a URL the user provides. Inspects HTML structure, CSS layout, typography, responsiveness, and JavaScript behavior; maps patterns to skills (html-css, javascript, react, next). Use when user pastes a link and asks how a site works, how it was made, reverse engineer a page, learn from a website, or recreate a layout. Voice triggers: "analyze this site", "how was this built", "break down this page", "teach me from this URL".

Maps the user's learning goal to an ordered DevLearn skill chain, suggested weekend project, and depth setting. Use when the user asks what to learn next, "I'm new where do I start", "I want a webpage/app/live site", or feels lost in the curriculum. Voice triggers: "what should I learn", "learning path".

Guided 60-second DevLearn setup. Asks persona, goal, depth, and stack; writes DEVLEARN.md, initializes .devlearn/ artifacts, runs install reminder, and gives a copy-paste first prompt. Use when user says "set up devlearn", "help me get started learning", first time using DevLearn, or DEVLEARN.md is missing. Voice triggers: "onboard", "devlearn setup", "get started with devlearn".

Ambient teaching layer for vibe coders and seasoned devs. After substantive changes, emits persona-aware lesson or decision blocks without blocking shipping. Use when user says "teach me while you work", "learn while I vibe", or when DEVLEARN.md exists with enabled true. Proactively apply after edit batches. Supports viber (What/Why/How) and seasoned (Decision/Alternatives/Risk) personas. Voice triggers: "teach me", "explain while you build", "explain like PR review".

Teaches APIs, REST, JSON, env vars, auth basics, and error handling while wiring real client-server or localStorage persistence. Use when the user wants to save data, fetch from a server, login, fix CORS, or curriculum routes here. Proactively suggest when adding fetch, routes, .env, or API handlers and DEVLEARN.md is enabled.

Pre-coding plain-English plan before large refactors or multi-file changes. Use when >5 files will change, architecture shifts, new dependencies, or user asks "what will you change before you do it". NOT the release checklist — use devlearn-pre-ship before merge/deploy. Pairs with devlearn-teach-while-coding when before_big_changes is true. Lifecycle: shared/ship-lifecycle.md. Voice triggers: "plan first", "before you refactor", "what are you about to change".

Teaches debugging while fixing real errors: reproduce, read stack/console, trace root cause, minimal fix, verify. Use when user pastes an error, "it's broken", blank page, failed test, or unexpected behavior. Proactively suggest when console/stack trace appears and DEVLEARN.md is enabled. Pairs with devlearn-apis (CORS/404), devlearn-react/next (hooks), devlearn-security (auth failures). Voice triggers: "debug this", "why is it broken".

Teaches deployment: build, hosting, environment on production, and what users actually hit when they open a URL. Use when the user wants to go live, share a link, set up CI, Docker, Vercel/Netlify/Fly, or curriculum routes here. Proactively suggest when adding deploy config and DEVLEARN.md is enabled.

Verify the connected Box MCP server. Confirms whether the user is on Box's official remote MCP at mcp.box.com (recommended; required for Box AI tools) or a community/deprecated Box MCP. Reports OAuth scopes granted and tool coverage. Surfaces harness-specific setup steps (Hermes, OpenClaw, Codex, Cursor, custom) when Box OAuth needs manual configuration. Use when setting up box-memory for the first time, when AI tools fail with tool-not-found errors, when OAuth fails with 401 or no-auth symptoms, or when the user asks which Box MCP they are using.

Create, update, or invoke a persistent Box AI Studio agent for memory recall. The "memory librarian" agent has locked instructions ("answer only from the indexed workspace, cite file IDs, never speculate"). Enterprise Advanced only. Use when the user wants a fixed-behavior AI assistant over their memory workspace, with locked guardrails instead of ad-hoc prompts.

Extract structured metadata from a binary file in Box using Box AI Extract Structured. Replaces "couldn't parse this format" stubs in companions with OCR + schema-driven extraction for PDFs / TIFF / PNG / JPEG. Returns extracted fields with confidence scores and source-text citations. Use during companion generation or when the user explicitly wants metadata extracted from a file.

Semantic Q&A over box-memory contents using Box AI. Wraps the Box AI /ai/ask endpoint - multi-doc Q&A across up to 25 files (Business+) or Hub-indexed Q&A across up to 20,000 files (Enterprise Plus). Returns answers with citations. Use when the user asks an open-ended question, box-recall returns sparse results, or when the user wants fuzzy semantic search instead of exact match.

Generate a paired companion markdown file for a binary in Box. The companion describes what the file is - summary, key facts, classification, related memories - pinned to a specific version via SHA256 hash. This is the no-chunking, no-embedding alternative to RAG. Use when the user uploads a binary and asks to remember what it is, describe a file, make a companion, or review a PDF/CAD/Office document.

Bootstrap a Box.com workspace for agent memory and file storage. Creates folder structure (or a Box Hub on Enterprise Plus), writes workspace config, detects tier and capabilities, defines the boxMemory metadata template (Business+) so Box AI Extract can target it. Use when the user wants to set up box-memory, initialize a workspace, or create a Box vault for the first time.

Find memories in a box-memory workspace by ID, wikilink, slug, structured filters, or free-text. Bypasses Box's 10-minute search indexing lag by reading the per-folder index files. Use when the user asks "what do I know about X", "what did we decide", "find memories tagged Y", "recall", or "search my Box memory".

Add or manage a team subtree in a box-memory workspace. Creates team folders with their own memories, files, and index - and tracks teams in workspace config. Box folder ACLs are the real isolation boundary; this skill creates structure only. Use when the user wants to create a team, list teams, inspect a team's contents, find cross-team conflicts, or remove a team.

Report the air-gap trust boundaries for the current session. Confirms Box Drive backend state, disk encryption posture, and identifies which LLM provider is handling agent prompts. Read-only. Use at the start of any sensitive workflow, or when an auditor asks what's closed vs open in the trust model.

Generate a paired companion markdown file for a binary in the local Box Drive workspace. Describes the file contents, key facts, hash-anchored to the current SHA256 of the binary. No chunking, no embedding - the binary stays whole, the companion describes it. FS-only, no Box API. Use when the user references a binary file (PDF, CAD, image, Office doc) and asks to describe, remember, or review it.

Probe the local Box Drive desktop-app mount to confirm the air-gapped backend is ready. Checks the mount path exists, is writable, Box Drive process is running, and reports the linked Box account email. Use when the user wants to verify Box Drive is set up, before running box-init for the first time, or when another skill needs the mount path. Never falls back to the network.

Rebuild a folder's _index.json (or the whole workspace's indexes) by walking the local Box Drive filesystem and re-reading every memory's frontmatter. Also resolves Box Drive .conflict files that arise when two clients write the same index, and detects stale companions whose binary changed. FS-only, no Box API.

Bootstrap a box-memory workspace on the local Box Drive mount. Creates folder structure, writes workspace config, seeds index files. FS-only - no Box API calls, no metadata template creation (the on-prem variant runs Personal-tier behavior; Business+ metadata templates are out of scope for air-gap mode). Use when the user wants to set up a Box-memory workspace under their local Box Drive folder.

Find memories in a local box-memory workspace by ID, wikilink, slug, structured filters, or free text. Reads only from local _index.json files - no Box Search, no Metadata Query API, no network. Use when the user asks what do I know about X, what did we decide, find memories tagged Y, recall, or search my Box memory.

Report the local box-memory workspace state - Box Drive backend health, teams, memory counts by kind/status, recent activity, index health. Read-only. Use when the user asks what's in their workspace, wants a status report, asks if Box Drive is syncing, or wants a sanity check before a heavy operation.

Manage team subtrees in a local box-memory workspace. Create, list, inspect, find cross-team conflicts, or remove a team from config. Folder ACLs are configured in Box (web UI or admin) - this skill creates filesystem structure only and cannot set permissions. FS-only, no Box API.

Add a pharma drug, herb, supplement, or biologic to the OpenMCAS curated compound library. Use when the user wants to extend data/compounds/seeds.json with a new compound. Triggers when the user says "add a compound", "I want to add X to the library", "include curcumin", or similar.

Scaffold a new experiments/EXP-NNN-*.md from the standard template and prefill it with the user's hypothesis, method, inputs, parameters, environment, and reproduction steps. Triggers when the user says "new experiment", "create an experiment report", "document this run", "EXP-006", or similar.

Capture a new MCAS trigger pattern (food, smell, physical, chemical, stress, infection, toxin, hormonal, medication) and add it to data/triggers/MCAS_Triggers_v1.csv with pathway mapping and linked counter-compounds. Triggers when the user says "report a trigger", "add this trigger", "I noticed when I'm around X I flare", or similar.

Run the full OpenMCAS computational pipeline end-to-end (build library, generate analogs, score warheads, screen targets, QSAR, rank) and produce a clean diff for review. Triggers when the user says "run the pipeline", "regenerate the rankings", "rerun everything", or similar.

Clear extension caches, analysis data, and drafts from the database. Use when testing sender lookups, analysis, or draft generation from a clean state.

Test the Electron app interactively using Chrome DevTools Protocol. Use when the user asks to test, verify, or interact with the running app via browser automation.

Pre-commit code review for production-critical issues. Use when reviewing staged changes, before committing, or when asked to review code for bugs and consistency issues.

Capture screenshots of the AOS Mail Electron app workflows using Playwright in demo mode. Use when the user asks for screenshots, workflow documentation, or visual captures of the app.

Use this skill whenever the user is operating on their Nakatomi CRM — creating, updating, finding, or deleting contacts, companies, deals, activities, notes, tasks, files, relationships. Trigger on mentions of "nakatomi", "the CRM", "my pipeline", "my deals", "add a contact", "log a call", "move a deal to …", or references to a Nakatomi workspace slug. Also trigger when the user asks to link a memory to a CRM entity or pull CRM context into their agent.

Use this skill when the user wants to audit or visually inspect their Nakatomi CRM. Trigger on phrases like "nakatomi dashboard", "open my CRM", "show me the dashboard", "audit the CRM", "what's in my pipeline right now", "boot the dashboard". This skill launches a local Nakatomi instance if one isn't running, enables the dashboard, and opens Chrome at http://localhost:8000/dashboard.

Configure VisionClaude channel settings — ElevenLabs API key, voice, and port

Pre-flight gate before editing code. Runs cheap on-device analysis first and returns high-signal findings with tight context packets, so the model fixes real problems instead of crawling the repo to rediscover them. Use BEFORE starting a fix/refactor/cleanup pass on a TypeScript project, or when asked to "clean up", "fix the errors", "lint", or "tighten up" a codebase.

Displays the alphabet of any language when the user asks for it, Use when the user requests the alphabet in a specific language or any language.