--- title: "Native skills + connectors: what improvements are possible" url: https://memory.wiki/LpLbS250 updated: 2026-05-15T16:43:23.665Z hub: https://memory.wiki/hub/raymindai bundle_count: 1 concept_count: 12 source: "strategy-note" --- # Native skills + connectors: what improvements are possible > Strategy note. What changes when we ship native skills / connectors on Claude, ChatGPT, and Gemini vs. the current URL-paste-only path. ## What changes at the "layer above AI" level | Today (URL / Chrome ext only) | With native skill / connector | |---|---| | User has to *remember* the URL and paste it | "Save this to my mdfy hub" as a single sentence | | Capture is Chrome-desktop-only | iOS / Android / Claude Desktop all covered | | Read-only (paste-as-context) | Bidirectional read + write — AI writes docs / updates bundles directly | | User has to *decide* "I should save this now" | End-of-conversation auto-capture (with consent) | | Hub search means going to mdfy.app | In-AI autocomplete: `@my-decisions`, `@cross-ai-thesis` | ## Per-platform menu ### 1. ChatGPT — `GPT + Actions` The most mature surface. OpenAPI schema is a single config file. - Wrap our public REST API (`/api/docs`, `/api/hub//recall`, `/api/bundles`) as actions - "mdfy memory" GPT — mid-conversation: *"what did I write about my cross-AI thesis?"* → action fires recall → answer cites the doc URL - End-of-conversation: *"Should I save this?"* → action POSTs `/api/docs` - Build cost: **near-zero**. OpenAPI spec + GPT instruction text. ### 2. Claude — `Claude Skill` + `Claude Code plugin` MCP server is already shipped (Desktop / Code / Cursor ✓). - **Claude.ai web Skill**: trigger `/mdfy` inside any Claude.ai conversation, including iOS Claude app — the surfaces our Chrome extension can't reach. - **Claude Code hooks package** (`mdfy-hooks` npm): pre-edit / post-edit / pre-commit hooks that auto-update `mdfy.app/d/` in real time. Same idea as the GitHub Action shipped this week, but at Claude Code session granularity. - Future: Anthropic **Computer Use** — experimental, Claude actually drives mdfy.app UI directly. ### 3. Gemini — `Gemini Extension` The largest potential user pool (Android default AI). - **Workspace fit**: Gmail → *"save this email as a memo"* → mdfy URL - **Docs fit**: Google Docs → *"cite from my mdfy hub"* → inline fetch + citation - Approval path is the heaviest of the three (Google review). ### 4. Cross-cutting (one-time build, three platforms benefit) - **iOS Share Sheet extension**: Safari / Mail / Slack → "Save to mdfy" - **macOS QuickLook plugin**: Finder space-bar on a `.md` URL → mdfy-rendered preview - **VS Code AI side panel deepening**: existing VS Code ext gets Cursor-style inline AI invocation ## Highest-leverage order (priority) 1. **ChatGPT GPT + Actions** — near-zero build cost, largest install base, public REST API already exists. Ship-able in 1-2 days. 2. **Claude Skill (.ai web)** — covers mobile Claude + the desktop Claude app that the Chrome ext can't reach. Available the moment Anthropic Skills API GA's. 3. **Claude Code hooks** — straight at the ICP. `mdfy-hooks` npm package. More real-time than the GitHub Action. 4. **iOS Share Sheet** — mobile capture is currently missing entirely. Apple review 1-2 weeks + Swift code ~1 week. ## Skip / defer (at least until post-v6 launch) - Gemini Extension — Google approval path is heavy + Android ICP fit is weak - NotebookLM connector — cannibalisation risk (NotebookLM is itself a hub product) - OpenAI Apps SDK — too new (stability uncertain) ## The single biggest synergy effect Shipping ChatGPT GPT + Claude Skill **simultaneously** turns the cross-AI thesis into a demonstrable artifact: > Save in ChatGPT → search from Claude immediately → 30-second video clip → goes into the Show HN post as evidence. That's the cross-AI moat made visible at the system level, not just at the URL-paste level. --- ## Summary Native skills and connectors on Claude, ChatGPT, and Gemini would enable bidirectional read/write access, mobile coverage, in-AI autocomplete search, and auto-capture of conversations, moving beyond the current URL-paste-only approach. The highest-leverage implementation prioritizes ChatGPT GPT+Actions (near-zero cost), Claude Skill for mobile coverage, and Claude Code hooks for real-time updates, with simultaneous shipping creating a demonstrable cross-AI workflow artifact. ## Themes - Native AI platform integrations - Cross-AI interoperability moat - Mobile capture gap closure ## Key takeaways - Native skills enable bidirectional read/write access whereas current URL-paste approach is read-only context insertion. - ChatGPT GPT+Actions is shippable in 1-2 days with near-zero build cost using existing public REST API and OpenAPI spec. - Claude Skill for claude.ai web covers mobile and desktop surfaces unreachable by the Chrome extension. - Claude Code hooks package allows real-time mdfy hub updates at session granularity through pre/post-edit hooks. - iOS Share Sheet extension addresses a current gap: mobile capture is entirely missing from current offerings. - Simultaneous ChatGPT and Claude Skill launch creates demonstrable cross-AI artifact that visualizes the competitive moat. ## Insights - The document frames native connectors not as feature parity but as unlocking bidirectional sync and real-time collaboration that URL-paste fundamentally cannot provide. - Shipping ChatGPT and Claude simultaneously creates a network effect that transforms an internal capability into externally visible competitive differentiation. - Platform prioritization is explicitly inverted from typical roadmaps: skip the largest potential user base (Gemini/Android) due to weak ICP fit and approval burden. ## Open questions / gaps - What specific technical constraints prevent the Gemini Extension from being viable despite Android's large install base? - How does the NotebookLM connector create cannialization risk if both are hub products? ## Concepts in this document - **Chrome extension** _(entity)_ Surface within memory.wiki ecosystem. - **Claude** _(entity)_ Specified AI tool for prototyping and validation before moving to high-fidelity design. - **MCP server** _(concept)_ A standard protocol allowing diverse AI tools to query and interact with the memory.wiki hub. - **ChatGPT** _(entity)_ One of the AI platforms currently suffering from isolated memory silos. - **Electron desktop app** _(entity)_ Current macOS/Windows distribution surface with high maintenance cost and limited unique value vs. web. - **Competitive Positioning** _(tag)_ Strategic differentiation from existing solutions in the AI memory and knowledge management space - **mdfy.app** _(entity)_ Legacy product being migrated to memory.wiki brand with 1+ year redirect maintenance planned. - **Authored, not extracted** _(concept)_ Principle that user intent comes first and AI linking is secondary, preventing product from reading as auto-organization tool. - **Claude Skill** _(entity)_ Web and desktop integration for Claude that reaches mobile and app surfaces the Chrome extension cannot cover. - **Graph in response** _(concept)_ Architecture where graph data lives in URL responses rather than behind service calls - **Platform Integration** _(tag)_ Category covering native AI platform skills and connectors implementation - **Cross-AI structural moat** _(concept)_ Competitive differentiation through one URL working across multiple AI platforms rather than fragmented per-AI variants. ## Concept relations (within this doc's concepts) - **mdfy.app** available via **Chrome extension** - **Claude Skill** enables web and mobile access **mdfy.app** - **Claude Skill** demonstrates strategy **Platform Integration** ## Bundles containing this document - [Phase 2 strategy: connectors, replacements, first-principles](https://memory.wiki/b/Ih0eUczw) > Three working notes from a strategy review session on 2026-05-15. Native AI integrations roadmap, what current surfaces to replace, first-principles drift check. _Hub canonical:_ https://memory.wiki/hub/raymindai _Concept digest:_ https://memory.wiki/raw/hub/raymindai?digest=1&compact=1