@framers/agentos-skills 0.1.0

package/README.md

@@ -0,0 +1,78 @@
+# @framers/agentos-skills
+
+Curated catalog of **skills** for the [AgentOS](https://github.com/framersai/agentos) ecosystem.
+
+[![npm](https://img.shields.io/npm/v/@framers/agentos-skills?logo=npm&color=cb3837)](https://www.npmjs.com/package/@framers/agentos-skills)
+
+```bash
+npm install @framers/agentos-skills
+```
+
+## What's Inside
+
+This package is **data only** — no runtime code, no heavy dependencies. It ships:
+
+| File                          | Description                                                          |
+| ----------------------------- | -------------------------------------------------------------------- |
+| `registry/curated/*/SKILL.md` | 16+ curated skill definitions (weather, github, slack, notion, etc.) |
+| `registry.json`               | Flat JSON index of every skill with metadata                         |
+| `types.d.ts`                  | TypeScript type declarations for the registry schema                 |
+
+Each skill is a directory containing a `SKILL.md` file with **YAML frontmatter** (metadata, requirements, install specs) and a **markdown body** (instructions injected into an agent's system prompt).
+
+## Available Skills
+
+| Category            | Skills                                                 |
+| ------------------- | ------------------------------------------------------ |
+| **Information**     | weather, summarize                                     |
+| **Developer Tools** | github, coding-agent                                   |
+| **Communication**   | slack-helper, discord-helper                           |
+| **Productivity**    | notion, obsidian, trello, apple-notes, apple-reminders |
+| **DevOps**          | healthcheck                                            |
+| **Media**           | spotify-player, whisper-transcribe                     |
+| **Security**        | 1password                                              |
+| **Creative**        | image-gen                                              |
+
+## Usage
+
+### Raw data (no dependencies needed)
+
+```typescript
+import registry from '@framers/agentos-skills';
+
+// registry.json is the full catalog
+console.log(registry.skills.curated.length); // 16+
+```
+
+### With the typed SDK
+
+For programmatic queries, filtering, and runtime loading, use [`@framers/agentos-skills-registry`](https://www.npmjs.com/package/@framers/agentos-skills-registry):
+
+```bash
+npm install @framers/agentos-skills-registry
+```
+
+```typescript
+import { searchSkills, getSkillsByCategory } from '@framers/agentos-skills-registry/catalog';
+
+const devTools = getSkillsByCategory('developer-tools');
+const matches = searchSkills('github');
+```
+
+## Relationship to Other Packages
+
+```
+@framers/agentos-skills          ← You are here (data: SKILL.md files + JSON index)
+  └── @framers/agentos-skills-registry   (SDK: typed catalog, query helpers, registry factories)
+        └── @framers/agentos             (optional peer: live SkillRegistry + snapshots)
+```
+
+| Package                              | What                                      | Runtime Code | Dependencies                           |
+| ------------------------------------ | ----------------------------------------- | :----------: | -------------------------------------- |
+| **@framers/agentos-skills**          | Raw SKILL.md files + JSON index           |      No      | Zero                                   |
+| **@framers/agentos-skills-registry** | Typed catalog + query helpers + factories |     Yes      | `agentos-skills`, optionally `agentos` |
+| **@framers/agentos**                 | Full cognitive runtime with SkillRegistry |     Yes      | Many                                   |
+
+## License
+
+MIT

package/package.json

@@ -0,0 +1,40 @@
+{
+  "name": "@framers/agentos-skills",
+  "version": "0.1.0",
+  "description": "Community registry of skills for AgentOS (SKILL.md prompt modules)",
+  "type": "module",
+  "sideEffects": false,
+  "exports": {
+    ".": {
+      "import": "./registry.json",
+      "types": "./types.d.ts"
+    },
+    "./registry.json": "./registry.json",
+    "./package.json": "./package.json"
+  },
+  "files": [
+    "registry",
+    "registry.json",
+    "types.d.ts",
+    "README.md"
+  ],
+  "publishConfig": {
+    "access": "public"
+  },
+  "keywords": [
+    "agentos",
+    "skills",
+    "registry",
+    "prompts"
+  ],
+  "author": {
+    "name": "Framers AI",
+    "email": "team@frame.dev",
+    "url": "https://frame.dev"
+  },
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/framersai/agentos-skills.git"
+  }
+}

package/registry/curated/1password/SKILL.md

@@ -0,0 +1,53 @@
+---
+name: 1password
+version: '1.0.0'
+description: Query and retrieve items from 1Password vaults using the 1Password CLI for secure credential access.
+author: Wunderland
+namespace: wunderland
+category: security
+tags: [1password, passwords, secrets, vault, credentials, security]
+requires_secrets: []
+requires_tools: []
+metadata:
+  agentos:
+    emoji: "\uD83D\uDD10"
+    primaryEnv: OP_SERVICE_ACCOUNT_TOKEN
+    homepage: https://developer.1password.com/docs/cli
+    requires:
+      bins: ['op']
+    install:
+      - id: brew
+        kind: brew
+        formula: 1password-cli
+        bins: ['op']
+        label: 'Install 1Password CLI (brew)'
+---
+
+# 1Password Vault Queries
+
+You can query 1Password vaults to retrieve credentials, secure notes, API keys, and other secrets using the `op` CLI. This enables secure access to stored credentials without hardcoding secrets in code or configuration files.
+
+When retrieving items, use `op item get` with the item name or UUID. Support querying specific fields (username, password, TOTP, custom fields) using the `--fields` flag. For listing items, filter by vault, category (login, secure-note, api-credential, credit-card), or tags. Always use the most specific identifier available to avoid ambiguous matches.
+
+For security, never display full passwords or secret values in plain text unless the user explicitly requests it. Instead, confirm the item exists and describe its metadata (title, vault, category, last modified). When injecting secrets into environment variables or configuration files, use `op run` or `op inject` for ephemeral secret injection that avoids writing secrets to disk.
+
+Support common workflows like looking up API keys for service integrations, retrieving database credentials for connection strings, and checking TOTP codes for two-factor authentication. When multiple items match a query, present a disambiguated list with vault and category context so the user can select the correct one.
+
+## Examples
+
+- "Look up the API key for our Stripe integration"
+- "What vaults do I have access to?"
+- "Get the database connection credentials from the Production vault"
+- "Generate a TOTP code for my AWS account"
+- "List all items tagged 'deploy' in the DevOps vault"
+- "Inject secrets from the 'staging-env' item into environment variables"
+
+## Constraints
+
+- Requires the `op` CLI to be installed and authenticated (service account token or interactive sign-in).
+- The agent can only access vaults and items that the authenticated account has permissions for.
+- Biometric unlock is not available in non-interactive CLI sessions; use service account tokens.
+- TOTP codes are time-sensitive and expire every 30 seconds.
+- Cannot create, modify, or delete vault items -- read-only access for security.
+- Session tokens expire after 30 minutes of inactivity by default.
+- Do not log, cache, or persist retrieved secret values beyond the immediate use.

package/registry/curated/apple-notes/SKILL.md

@@ -0,0 +1,45 @@
+---
+name: apple-notes
+version: '1.0.0'
+description: Create, read, search, and manage notes in Apple Notes using AppleScript and macOS automation.
+author: Wunderland
+namespace: wunderland
+category: productivity
+tags: [apple-notes, macos, notes, applescript, automation]
+requires_secrets: []
+requires_tools: [filesystem]
+metadata:
+  agentos:
+    emoji: "\uD83D\uDCF1"
+    os: ['darwin']
+    requires:
+      bins: ['osascript']
+---
+
+# Apple Notes Integration
+
+You can create, read, search, and manage notes in Apple Notes on macOS using AppleScript commands executed via `osascript`. This provides native access to the Notes app without requiring third-party tools or API keys.
+
+When creating notes, use `osascript` to invoke AppleScript that creates note entries with title and body text. Notes support basic HTML formatting for bold, italic, lists, and headings. Organize notes into folders by specifying the target folder during creation. If a folder does not exist, create it first before adding notes to it.
+
+For reading and searching notes, query the Notes app for notes matching specific titles, body content, or folder locations. When listing notes, include the note title, creation date, modification date, and folder. For search operations, match against both title and body text. Present results sorted by modification date (most recent first) by default.
+
+When modifying existing notes, append content to the end of the note body unless the user specifies a different location. Avoid overwriting existing note content without explicit confirmation. Support bulk operations like moving multiple notes between folders or exporting note contents to Markdown files on the filesystem.
+
+## Examples
+
+- "Create a new note titled 'Meeting Notes - Feb 7' in my Work folder"
+- "Search my notes for anything about 'quarterly review'"
+- "List all notes in the Ideas folder sorted by date"
+- "Append today's action items to my 'Running Tasks' note"
+- "Export all notes from the Research folder to Markdown files"
+
+## Constraints
+
+- macOS only -- requires the Apple Notes app and `osascript` binary.
+- AppleScript access to Notes may require accessibility permissions to be granted.
+- Rich formatting is limited to basic HTML supported by Notes (bold, italic, lists, links).
+- Embedded images and attachments in notes cannot be read or created via AppleScript.
+- iCloud-synced notes are accessible but sync timing is controlled by the system.
+- Large notes (100KB+) may experience slow AppleScript operations.
+- Notes in locked/password-protected folders cannot be accessed via automation.

package/registry/curated/apple-reminders/SKILL.md

@@ -0,0 +1,46 @@
+---
+name: apple-reminders
+version: '1.0.0'
+description: Create, manage, and query reminders and lists in Apple Reminders using AppleScript and macOS automation.
+author: Wunderland
+namespace: wunderland
+category: productivity
+tags: [apple-reminders, macos, reminders, tasks, applescript, automation]
+requires_secrets: []
+requires_tools: [filesystem]
+metadata:
+  agentos:
+    emoji: "\u2705"
+    os: ['darwin']
+    requires:
+      bins: ['osascript']
+---
+
+# Apple Reminders Integration
+
+You can create, manage, complete, and query reminders in Apple Reminders on macOS using AppleScript commands executed via `osascript`. This provides native access to the Reminders app for task management without external services.
+
+When creating reminders, specify the title, due date, priority (0=none, 1-4=high, 5=medium, 6-9=low), and target list. If the user does not specify a list, use the default reminders list. Support creating reminders with notes/body text for additional context. For recurring reminders, set the recurrence rule (daily, weekly, monthly) when supported.
+
+For querying reminders, list items by list name, completion status, due date range, or priority level. Present results in a clear format showing title, due date, priority, and completion status. Support filtering incomplete reminders that are overdue. When showing upcoming reminders, sort by due date ascending.
+
+When managing reminders, support completing items (marking as done), updating due dates, changing priorities, moving between lists, and deleting items. For bulk operations, process items efficiently and report results. Create new reminder lists when the user requests organizing tasks into new categories.
+
+## Examples
+
+- "Create a reminder to 'Call dentist' tomorrow at 2pm in my Personal list"
+- "Show me all overdue reminders"
+- "Mark the 'Submit report' reminder as complete"
+- "List all reminders due this week sorted by priority"
+- "Create a new list called 'Home Renovation' and add 5 tasks to it"
+- "Move all high-priority reminders from Inbox to the Urgent list"
+
+## Constraints
+
+- macOS only -- requires the Apple Reminders app and `osascript` binary.
+- AppleScript access to Reminders may require accessibility permissions.
+- Location-based reminders cannot be created via AppleScript.
+- Sub-tasks (nested reminders) have limited AppleScript support.
+- iCloud-synced reminders are accessible but sync timing is system-controlled.
+- Attachments and images on reminders cannot be managed via automation.
+- Completed reminders may be automatically hidden based on the user's Reminders app settings.

package/registry/curated/coding-agent/SKILL.md

@@ -0,0 +1,40 @@
+---
+name: coding-agent
+version: '1.0.0'
+description: Write, review, debug, refactor, and explain code across multiple programming languages and frameworks.
+author: Wunderland
+namespace: wunderland
+category: developer-tools
+tags: [coding, programming, code-review, debugging, refactoring, development]
+requires_secrets: []
+requires_tools: [filesystem]
+metadata:
+  agentos:
+    emoji: "\uD83D\uDCBB"
+---
+
+# Code Writing and Review Agent
+
+You are a skilled software developer capable of writing, reviewing, debugging, refactoring, and explaining code across a wide range of programming languages and frameworks. Approach every coding task with attention to correctness, readability, maintainability, and performance.
+
+When writing code, always follow the conventions of the target language and framework. Use consistent naming conventions, add appropriate error handling, and include inline comments for non-obvious logic. For new features, write modular, testable code with clear separation of concerns. Suggest and write unit tests alongside implementation code when appropriate.
+
+For code review, analyze code for bugs, security vulnerabilities, performance issues, and style inconsistencies. Provide specific, actionable feedback with line references and suggested fixes. Prioritize issues by severity: security vulnerabilities first, then correctness bugs, then performance, then style. Explain the reasoning behind each suggestion.
+
+When debugging, systematically narrow down the root cause by analyzing error messages, stack traces, and code flow. Suggest targeted debugging strategies (logging, breakpoints, bisection) rather than shotgun approaches. For refactoring, preserve existing behavior while improving code structure, and recommend incremental changes over big-bang rewrites.
+
+## Examples
+
+- "Write a TypeScript function to debounce API calls with a configurable delay"
+- "Review this PR for security issues and suggest improvements"
+- "Debug why this React component re-renders on every keystroke"
+- "Refactor this 200-line function into smaller, testable units"
+- "Explain how this recursive algorithm works step by step"
+
+## Constraints
+
+- Always test suggestions mentally for edge cases before presenting them.
+- Do not introduce dependencies without explaining why they are necessary.
+- Respect existing code style and conventions in the project.
+- For security-sensitive code (auth, crypto, input validation), err on the side of caution and recommend established libraries over custom implementations.
+- Large refactors should be broken into reviewable increments.

package/registry/curated/discord-helper/SKILL.md

@@ -0,0 +1,43 @@
+---
+name: discord-helper
+version: '1.0.0'
+description: Manage Discord servers, channels, roles, and messages through the Discord API.
+author: Wunderland
+namespace: wunderland
+category: communication
+tags: [discord, messaging, server, moderation, community]
+requires_secrets: [discord.bot_token]
+requires_tools: []
+metadata:
+  agentos:
+    emoji: "\uD83C\uDFAE"
+    primaryEnv: DISCORD_BOT_TOKEN
+    homepage: https://discord.com/developers
+---
+
+# Discord Server Helper
+
+You can interact with Discord servers (guilds) to send messages, manage channels, assign roles, moderate content, and handle events. Use the Discord API with the configured bot token to perform server management tasks.
+
+When sending messages, use Discord's rich embed format for structured content including titles, descriptions, fields, colors, and thumbnails. Respect channel categories and permissions when posting. For moderation tasks, always log actions and provide clear reasons when warning, muting, or banning users.
+
+For server management, you can create and organize channels into categories, set up role hierarchies with appropriate permissions, and configure server settings. When handling voice channels, check user presence before attempting operations. Use threads for focused discussions to keep channels organized.
+
+Manage slash commands and interactions for interactive bot experiences. Handle reaction roles, welcome messages, and automated moderation rules. When dealing with large servers, paginate member lists and message histories efficiently. Always respect Discord's rate limits and retry with exponential backoff when throttled.
+
+## Examples
+
+- "Send an embed to #announcements with the release notes"
+- "Create a new text channel called 'dev-chat' in the Engineering category"
+- "List all members with the 'Moderator' role"
+- "Set up a reaction role message in #roles for team selection"
+- "Purge the last 50 messages in #spam"
+
+## Constraints
+
+- Bot permissions are determined by its role in each server. Common permissions needed: Send Messages, Manage Channels, Manage Roles, Kick/Ban Members.
+- Rate limits: 50 requests/second globally, with per-route limits.
+- Message content intent is required for reading message content in non-DM contexts.
+- Cannot interact with servers the bot has not been invited to.
+- Embeds are limited to 6,000 total characters across all fields.
+- Bulk message deletion only works for messages less than 14 days old.

package/registry/curated/git/SKILL.md

@@ -0,0 +1,32 @@
+---
+name: git
+description: Work with Git repositories from the command line.
+metadata:
+  agentos:
+    emoji: '🧰'
+    requires:
+      bins: ['git']
+    install:
+      - id: brew
+        kind: brew
+        formula: git
+        bins: ['git']
+        label: 'Install git (brew)'
+      - id: apt
+        kind: apt
+        package: git
+        bins: ['git']
+        os: ['linux']
+        label: 'Install git (apt)'
+---
+
+# Git
+
+Use `git` to inspect history, create branches, commit changes, and resolve conflicts.
+
+## Common workflows
+
+- Check status: `git status`
+- Create a branch: `git checkout -b my-branch`
+- Stage + commit: `git add -A && git commit -m "message"`
+- Rebase: `git rebase -i origin/main`

package/registry/curated/github/SKILL.md

@@ -0,0 +1,54 @@
+---
+name: github
+version: '1.0.0'
+description: Manage GitHub repositories, issues, pull requests, releases, and Actions workflows using the gh CLI.
+author: Wunderland
+namespace: wunderland
+category: developer-tools
+tags: [github, git, issues, pull-requests, ci-cd, code-review]
+requires_secrets: [github.token]
+requires_tools: []
+metadata:
+  agentos:
+    emoji: "\uD83D\uDC19"
+    primaryEnv: GITHUB_TOKEN
+    requires:
+      bins: ['gh']
+    install:
+      - id: brew
+        kind: brew
+        formula: gh
+        bins: ['gh']
+        label: 'Install GitHub CLI (brew)'
+      - id: apt
+        kind: apt
+        package: gh
+        bins: ['gh']
+        os: ['linux']
+        label: 'Install GitHub CLI (apt)'
+---
+
+# GitHub (gh CLI)
+
+Use the `gh` CLI to interact with GitHub repositories, issues, pull requests, releases, and GitHub Actions. You have full access to the GitHub API through the CLI, which supports both interactive and scriptable workflows.
+
+When managing issues, always check for existing duplicates before creating new ones. For pull requests, include a clear title and description summarizing the changes. When reviewing PRs, provide specific, actionable feedback referencing line numbers. Use labels and milestones to organize work when the repository supports them.
+
+For repository operations, prefer `gh api` for advanced queries that the standard subcommands do not cover. You can use `gh api graphql` for complex queries involving nested relationships. Always verify authentication status with `gh auth status` before performing write operations.
+
+When working with GitHub Actions, you can trigger workflows with `gh workflow run`, check run status with `gh run list`, and view logs with `gh run view --log`. Use `gh release create` to manage releases with proper semantic versioning and changelogs.
+
+## Examples
+
+- `gh issue list --label bug --state open`
+- `gh pr create --title "Fix auth bug" --body "Resolves #42"`
+- `gh pr review 123 --approve`
+- `gh api repos/{owner}/{repo}/actions/runs --jq '.workflow_runs[:5]'`
+- `gh release create v1.2.0 --generate-notes`
+
+## Constraints
+
+- Requires the `gh` CLI to be installed and authenticated.
+- Write operations require appropriate repository permissions.
+- API rate limits apply (5,000 requests/hour for authenticated users).
+- Large file operations should use Git LFS rather than the GitHub API.

package/registry/curated/healthcheck/SKILL.md

@@ -0,0 +1,43 @@
+---
+name: healthcheck
+version: '1.0.0'
+description: Monitor health and availability of systems, services, APIs, and infrastructure endpoints.
+author: Wunderland
+namespace: wunderland
+category: devops
+tags: [monitoring, health, uptime, infrastructure, diagnostics, status]
+requires_secrets: []
+requires_tools: [web-search]
+metadata:
+  agentos:
+    emoji: "\uD83C\uDFE5"
+    requires:
+      anyBins: ['curl', 'wget']
+---
+
+# System and Service Health Monitoring
+
+You can check the health and availability of systems, services, APIs, and infrastructure endpoints. Use HTTP requests, ping commands, and service-specific health check protocols to assess operational status.
+
+When performing health checks, test multiple dimensions: HTTP response codes, response times, SSL certificate validity, DNS resolution, and content correctness. For API endpoints, verify not just that they respond with 200 OK, but that the response body matches expected schemas. Track response latency and flag anything over reasonable thresholds (e.g., >2s for web pages, >500ms for API calls).
+
+For infrastructure monitoring, check CPU, memory, disk usage, and network connectivity when system-level access is available. Aggregate results into a clear status dashboard format: green (healthy), yellow (degraded), red (down). Always include timestamps with health check results for audit trails.
+
+When diagnosing failures, follow a systematic approach: check DNS resolution first, then TCP connectivity, then TLS handshake, then HTTP response. For intermittent issues, suggest monitoring intervals and alert thresholds. Provide remediation suggestions alongside failure reports when possible.
+
+## Examples
+
+- "Check if api.example.com is responding and measure latency"
+- "Verify the SSL certificate for example.com is valid and not expiring soon"
+- "Run health checks on all our microservice endpoints"
+- "Check the status of our database, Redis, and message queue connections"
+- "Monitor the /health endpoint every 30 seconds and alert on failures"
+
+## Constraints
+
+- Can only check endpoints that are network-accessible from the agent's environment.
+- Deep application-level health checks require appropriate authentication and access.
+- Cannot install monitoring agents or modify system configurations.
+- Rate-limited health checks should respect the target service's acceptable request frequency.
+- SSL certificate checks are informational; the agent cannot renew or modify certificates.
+- Internal/private network services may not be reachable depending on the agent's network context.

package/registry/curated/image-gen/SKILL.md

@@ -0,0 +1,44 @@
+---
+name: image-gen
+version: '1.0.0'
+description: Generate images from text prompts using AI image generation APIs like DALL-E, Stable Diffusion, or Midjourney.
+author: Wunderland
+namespace: wunderland
+category: creative
+tags: [image-generation, ai-art, dall-e, stable-diffusion, creative, visual]
+requires_secrets: [openai.api_key]
+requires_tools: []
+metadata:
+  agentos:
+    emoji: "\uD83C\uDFA8"
+    primaryEnv: OPENAI_API_KEY
+    homepage: https://platform.openai.com/docs/guides/images
+---
+
+# AI Image Generation
+
+You can generate images from text descriptions using AI image generation APIs. Craft detailed, effective prompts that translate the user's creative vision into high-quality generated images.
+
+When generating images, help the user refine their prompt for best results. A good image prompt includes: subject description, style (photorealistic, illustration, watercolor, etc.), composition (close-up, wide shot, overhead), lighting (natural, dramatic, soft), color palette, and mood/atmosphere. Offer prompt suggestions when the user's description is vague or underspecified.
+
+Support different image sizes and aspect ratios based on the API capabilities (1024x1024, 1792x1024, 1024x1792 for DALL-E 3). For iterative refinement, maintain context from previous generations so the user can say "make it more vibrant" or "change the background to a beach." Save generated images to the filesystem when the user requests it, with descriptive filenames.
+
+When the user requests variations or edits of existing images, use the appropriate API endpoints (variations, inpainting) when available. For batch generation, create multiple variations with slightly different prompts to give the user options. Always inform the user of the model and settings used for each generation.
+
+## Examples
+
+- "Generate an image of a cozy cabin in the mountains at sunset, watercolor style"
+- "Create a professional logo for a coffee shop called 'Bean There'"
+- "Make the previous image more dramatic with storm clouds"
+- "Generate 3 variations of a cyberpunk cityscape at night"
+- "Create a 16:9 landscape of a serene Japanese garden in spring"
+
+## Constraints
+
+- Image generation costs API credits per request; inform the user of approximate costs when possible.
+- Content policy restrictions apply: no realistic faces of real people, no violent/explicit content.
+- DALL-E 3 does not support exact image editing or inpainting; describe the full desired output.
+- Generated images may not perfectly match the prompt; iterative refinement is expected.
+- Maximum prompt length varies by model (DALL-E 3: 4,000 characters).
+- Image quality and style depend on the model version and generation parameters.
+- Generated images should not be represented as photographs or real events.

package/registry/curated/notion/SKILL.md

@@ -0,0 +1,43 @@
+---
+name: notion
+version: '1.0.0'
+description: Read, create, and manage pages, databases, and content blocks in Notion workspaces.
+author: Wunderland
+namespace: wunderland
+category: productivity
+tags: [notion, wiki, database, notes, project-management, knowledge-base]
+requires_secrets: [notion.api_key]
+requires_tools: []
+metadata:
+  agentos:
+    emoji: "\uD83D\uDCD3"
+    primaryEnv: NOTION_API_KEY
+    homepage: https://developers.notion.com
+---
+
+# Notion Workspace
+
+You can interact with Notion workspaces to create, read, update, and search pages and databases. Use the Notion API to manage content blocks, database entries, and page properties programmatically.
+
+When creating pages, structure content using Notion's block types: paragraphs, headings (h1/h2/h3), bulleted lists, numbered lists, to-do items, code blocks, callouts, and toggle blocks. Always use appropriate heading hierarchy for document structure. For databases, define property schemas with the correct types (title, rich_text, number, select, multi_select, date, checkbox, url, email, phone, formula, relation, rollup).
+
+For search operations, use the Notion search endpoint with query text and optional filters by object type (page or database). When updating existing pages, preserve the existing block structure and only modify the specific blocks that need changes. Append new content at the end unless the user specifies a different location.
+
+When working with database views, respect existing filters and sorts. Create new database entries with all required properties filled in. For relational databases, verify that referenced pages exist before creating relations. Handle pagination for large result sets by following cursor-based pagination tokens.
+
+## Examples
+
+- "Create a new page in my Project Notes database with title 'Q1 Planning'"
+- "Search my workspace for pages about 'onboarding'"
+- "Add a to-do list to the meeting notes page with action items from the standup"
+- "Query the Tasks database for all items assigned to me that are in progress"
+- "Update the status of task #42 to 'Complete'"
+
+## Constraints
+
+- API rate limit: 3 requests/second per integration.
+- Page content is limited to 100 blocks per append operation.
+- Rich text segments are limited to 2,000 characters each.
+- The integration can only access pages and databases explicitly shared with it.
+- Nested blocks (children of children) require separate API calls to retrieve.
+- File and media blocks cannot be created via API; only existing file URLs can be embedded.

package/registry/curated/obsidian/SKILL.md

@@ -0,0 +1,42 @@
+---
+name: obsidian
+version: '1.0.0'
+description: Read, create, and manage notes, links, and metadata in Obsidian vaults via the local filesystem.
+author: Wunderland
+namespace: wunderland
+category: productivity
+tags: [obsidian, markdown, notes, knowledge-graph, zettelkasten, pkm]
+requires_secrets: []
+requires_tools: [filesystem]
+metadata:
+  agentos:
+    emoji: "\uD83D\uDC8E"
+    homepage: https://obsidian.md
+---
+
+# Obsidian Vault Interaction
+
+You can interact with Obsidian vaults by reading and writing Markdown files directly on the local filesystem. Obsidian vaults are simply directories of `.md` files with optional YAML frontmatter and `[[wikilink]]` syntax for inter-note linking.
+
+When creating new notes, always include YAML frontmatter with relevant metadata fields like `tags`, `date`, `aliases`, and any custom properties the vault uses. Use `[[wikilinks]]` for internal links and `![[embeds]]` for transclusion. Respect the vault's folder structure -- check for existing organizational patterns (e.g., daily notes in `Daily/`, templates in `Templates/`) before creating files in new locations.
+
+For searching and navigating the vault, scan file contents for keywords, tags (`#tag` syntax), and frontmatter properties. Follow `[[wikilinks]]` to traverse the knowledge graph. When summarizing vault contents, consider both the explicit folder hierarchy and the implicit link-based graph structure.
+
+When editing existing notes, preserve all existing frontmatter fields, wikilinks, and formatting. Append new content at appropriate locations rather than overwriting. For daily notes, follow the vault's date format convention (typically `YYYY-MM-DD`). Support Dataview-compatible frontmatter when the user's vault uses the Dataview plugin.
+
+## Examples
+
+- "Create a new note called 'Project Kickoff' in the Meetings folder with today's date"
+- "Find all notes tagged #research and summarize their key points"
+- "Add a link to [[Architecture Decisions]] in the project overview note"
+- "List all notes that link to [[API Design]] (backlinks)"
+- "Create a daily note for today with the standup template"
+
+## Constraints
+
+- Operates on local filesystem only; no cloud sync awareness.
+- Cannot interact with Obsidian plugins directly (Canvas, Excalidraw, etc.) -- only reads/writes Markdown files.
+- Binary attachments (images, PDFs) can be referenced but not created.
+- Vault path must be known and accessible to the agent.
+- Wikilink resolution follows Obsidian's "shortest path" convention when note names are unique.
+- Large vaults (10,000+ notes) may require targeted searches rather than full scans.