Apps are optional tool domains that extend Nexus with third-party integrations and desktop workflows. Each app brings its own tools, credentials, and API connections; enable only what you need.
Configure apps in Settings -> Nexus -> Apps. Install an app, enter your API key when required, hit Validate, then toggle it on. Built-in apps do not download extra npm packages when enabled.
Experimental: Composer, Web Tools, Nexus Ingester, and Data Analysis are new and may have rough edges. Please report issues if you run into problems.
| App | Tools | What It Does |
|---|---|---|
| Skills | listSkills, loadSkill, createSkill, updateSkill, archiveSkill, syncSkills | Author, index, and load reusable agent Skills (per the Skills Protocol) straight from your vault. Discover available skills, load one with its bundled files into context, and create/update/archive them over time. A background watcher keeps the index in sync as you edit skill files. Skills are vault-local — Nexus never reaches into OS-home provider folders. Works on desktop and mobile. |
| Data Analysis (experimental, desktop only) | runPython, listCapabilities | Run Python (pandas) against your vault's CSV and Excel data in a sandboxed Pyodide worker. Workbooks project into editable CSVs and write back automatically — formulas, charts, images, and pivots are preserved byte-for-byte. No setup beyond enabling the app; the Python engine loads on first use. Requires desktop; not available on mobile. |
| ElevenLabs | textToSpeech, listVoices, soundEffects, generateMusic | AI audio generation: convert text to speech, create sound effects, and generate music. Audio files save directly to your vault. When enabled, ElevenLabs voices also become available to the built-in read-aloud and generateAudio flows through the Voice defaults. |
| Nexus Ingester (experimental) | ingest, listCapabilities | Convert PDF, DOCX, PPTX, and audio files in your vault to sibling Markdown notes. Two modes: Manual - right-click any supported file and choose "Convert to Markdown". Auto - enable "Auto-convert new files" in Settings -> Defaults -> Ingestion and any supported file added to the vault is converted automatically. PDF extraction uses text mode or OCR. OCR supports Mistral OCR (native, or via OpenRouter) — which returns full-fidelity text and extracts embedded images into a per-note folder, linked inline — as well as vision-model OCR. The PDF worker is loaded only when PDF ingestion runs. Audio transcription supports OpenAI (Whisper, GPT-4o Transcribe), Groq (Whisper), and Google Gemini multimodal audio. XLSX ingestion is not included in the core release. |
| Composer (experimental) | compose, listFormats | Combine multiple files into one. Merge PDFs, concatenate Markdown files, or mix and concat audio tracks. Audio input supports common browser-decodable formats such as MP3 and WAV. Audio output supports WAV and WebM/Opus. Supports per-track volume, offset, and fade controls for audio mixing. |
| Web Tools (experimental, desktop only) | openWebpage, capturePagePdf, capturePagePng, captureToMarkdown, extractLinks | Open any webpage in a headless browser and capture it as a PDF, PNG, or clean Markdown (boilerplate stripped). Also extracts all links with their text and type. Requires desktop; not available on mobile. |
Have an idea for a new app? Open an issue with the app-request label.
Want to build your own? See Building Apps, an agentic prompt you can feed to your AI coding assistant to create a new app from scratch. It covers the full pattern: manifest, agent class, tools, vault file saving, and registration.