skillli 0.1.0

package/.claude/skills/skillli/SKILL.md

@@ -0,0 +1,71 @@
+---
+name: skillli
+description: >
+  Discover, search, install, rate, and manage agentic AI skills from the
+  skillli registry. Use this skill when the user wants to find skills for
+  a specific task, install a skill, publish their own skill, check skill
+  ratings, or when Claude needs to search for relevant skills and context
+  packages to accomplish a task. Triggers on: /skillli, "find a skill",
+  "search for skills", "install skill", "skill marketplace".
+disable-model-invocation: false
+user-invocable: true
+---
+
+# Skillli -- Agentic Skills Discovery and Management
+
+You are now acting as the Skillli skill librarian. Your role is to help
+users discover, evaluate, install, and manage agentic AI skills.
+
+## Available Actions
+
+### Search for Skills
+When the user needs a skill for a specific task:
+1. Use the skillli MCP server's `search_skills` tool with relevant query terms
+2. Present results in a clear table format with name, description, rating, and trust level
+3. Recommend the top result with reasoning
+
+### Trawl for Skills (Deep Search)
+When a simple search is insufficient, use broader discovery:
+1. Use the `trawl_skills` MCP tool for broad multi-source discovery
+2. The trawler searches: skillli registry, GitHub repos, npm packages
+3. Results are ranked by relevance and confidence
+4. Present findings with source attribution
+
+### Install a Skill
+When the user wants to install a skill:
+1. Use `get_skill_info` to show full details and trust score
+2. If trust score < 50, warn the user with specific concerns
+3. Use `install_skill` with link=true to install and auto-link to Claude Code
+
+### Rate a Skill
+After using a skill, prompt the user to rate it:
+1. Use `rate_skill` with the skill name and 1-5 rating
+2. Encourage adding a comment for community benefit
+
+### List Installed Skills
+Use the `installed-skills` resource to show what is currently installed.
+
+### Publish a Skill
+Guide users through creating and publishing skills:
+1. Run `skillli init` to create from template
+2. Edit the SKILL.md with instructions
+3. Run `skillli publish --dry-run` to validate
+4. Submit to the registry
+
+## Trawler Sub-Agent Pattern
+When you need to search broadly for skills:
+- Describe the search task clearly
+- Use the Task tool to spawn a research sub-agent
+- The sub-agent should use the trawl_skills MCP tool
+- Collect and synthesize results back to the user
+
+## Output Formatting
+- Use tables for search results
+- Use star ratings for display
+- Color-code trust levels: [OFFICIAL], [VERIFIED], [COMMUNITY]
+- Always show the trust score when recommending installation
+
+## Safety Reminders
+- Never install skills with trust score below 30 without explicit user confirmation
+- Always show safeguard warnings before installation
+- Recommend reviewing SKILL.md content before installing community skills

package/.claude/skills/skillli/references/skill-format-spec.md

@@ -0,0 +1,62 @@
+# Skillli Skill Format Specification
+
+## SKILL.md Structure
+
+Every skill package MUST contain a `SKILL.md` file with YAML frontmatter.
+
+### Required Fields
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `name` | string | Lowercase alphanumeric with hyphens (e.g. `my-skill`) |
+| `version` | string | Semver format (e.g. `1.0.0`) |
+| `description` | string | 10-500 characters describing the skill |
+| `author` | string | GitHub username or org |
+| `license` | string | SPDX license identifier (e.g. `MIT`) |
+| `tags` | string[] | 1-20 tags for discovery |
+| `category` | enum | One of: `development`, `creative`, `enterprise`, `data`, `devops`, `other` |
+
+### Optional Fields
+
+| Field | Type | Default | Description |
+|-------|------|---------|-------------|
+| `repository` | string | - | GitHub repo URL |
+| `homepage` | string | - | Homepage URL |
+| `min-skillli-version` | string | - | Minimum skillli version required |
+| `trust-level` | enum | `community` | `community`, `verified`, or `official` |
+| `checksum` | string | - | SHA-256 checksum (auto-generated on publish) |
+| `disable-model-invocation` | boolean | `false` | Prevent AI from auto-invoking |
+| `user-invocable` | boolean | `true` | Allow manual invocation via slash command |
+
+### Trust Levels
+
+- **community**: Published by any user; basic validation only
+- **verified**: Author identity verified; additional review
+- **official**: Maintained by the skillli team
+
+### Markdown Body
+
+The body after the frontmatter contains the skill instructions that Claude
+follows when the skill is invoked. Keep under 500 lines.
+
+## Skill Package Directory
+
+```
+my-skill/
+├── SKILL.md         # Required
+├── skillli.json     # Auto-generated manifest
+├── scripts/         # Optional helper scripts (.sh, .py, .js only)
+├── references/      # Optional reference documents
+├── assets/          # Optional templates, configs
+├── agents.md        # Optional agent instructions (CLAUDE.md equivalent)
+└── LICENSE          # Recommended
+```
+
+## Categories
+
+- **development**: Code, testing, CI/CD, debugging
+- **creative**: Writing, design, content creation
+- **enterprise**: Business, compliance, reporting
+- **data**: Analytics, ETL, visualization
+- **devops**: Infrastructure, deployment, monitoring
+- **other**: Everything else

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Domusgpt
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,147 @@
+# skillli
+
+**The Skill Librarian** — Discover, publish, rate, and use agentic AI skills packages.
+
+```
+npm install -g skillli
+```
+
+skillli is an npm package that provides a complete ecosystem for managing agentic AI skills (SKILL.md files). It includes a CLI, a core library, an MCP server for Claude Code integration, and an agentic trawler for multi-source skill discovery.
+
+## Features
+
+- **Skill Format Spec** — Standard SKILL.md format with YAML frontmatter for defining AI agent skills
+- **CLI** — `skillli init`, `search`, `install`, `publish`, `rate`, `trawl`, and more
+- **MCP Server** — Native Claude Code integration via Model Context Protocol
+- **Agentic Trawler** — Search across the registry, GitHub, and npm for relevant skills
+- **Safeguards** — Trust scoring, prohibited pattern detection, size limits, checksum verification
+- **Ratings** — Community ratings and reviews for skill quality
+
+## Quick Start
+
+```bash
+# Install globally
+npm install -g skillli
+
+# Create a new skill
+skillli init my-skill
+
+# Search for skills
+skillli search "code review"
+
+# Install a skill
+skillli install code-reviewer --link
+
+# Trawl across multiple sources
+skillli trawl "kubernetes deployment"
+```
+
+## CLI Commands
+
+| Command | Description |
+|---------|-------------|
+| `skillli init [name]` | Create a new skill from template |
+| `skillli search <query>` | Search the registry |
+| `skillli install <skill>` | Install a skill |
+| `skillli uninstall <skill>` | Uninstall a skill |
+| `skillli info <skill>` | Show skill details |
+| `skillli list` | List installed skills |
+| `skillli rate <skill> <1-5>` | Rate a skill |
+| `skillli update` | Sync registry index |
+| `skillli publish [path]` | Publish a skill |
+| `skillli trawl <query>` | Agentic multi-source search |
+
+## MCP Server
+
+skillli includes a built-in MCP server for Claude Code integration:
+
+```json
+{
+  "mcpServers": {
+    "skillli": {
+      "command": "npx",
+      "args": ["skillli-mcp"]
+    }
+  }
+}
+```
+
+### MCP Tools
+
+- `search_skills` — Search the registry by query, tags, or category
+- `install_skill` — Install and link a skill to .claude/skills/
+- `get_skill_info` — Get detailed skill metadata and trust score
+- `trawl_skills` — Multi-source agentic search
+- `rate_skill` — Submit a 1-5 star rating
+
+### MCP Resources
+
+- `skillli://installed` — List of installed skills
+- `skillli://index` — Full registry index
+
+## Skill Format
+
+Skills are defined as SKILL.md files with YAML frontmatter:
+
+```yaml
+---
+name: my-skill
+version: 1.0.0
+description: A clear description of what this skill does
+author: your-github-username
+license: MIT
+tags: [typescript, testing]
+category: development
+trust-level: community
+user-invocable: true
+---
+
+# My Skill
+
+Instructions that Claude follows when this skill is invoked...
+```
+
+### Categories
+
+`development` | `creative` | `enterprise` | `data` | `devops` | `other`
+
+### Trust Levels
+
+- **community** — Published by any user, basic validation
+- **verified** — Author identity verified, additional review
+- **official** — Maintained by the skillli team
+
+## Safeguards
+
+skillli runs safety checks on all skills:
+
+1. Schema validation (Zod)
+2. File size limits (5MB total, 500 lines for SKILL.md)
+3. Prohibited pattern scanning (eval, exec, credentials, etc.)
+4. Script allowlisting (.sh, .py, .js only)
+5. Checksum verification
+6. Trust score computation (0-100)
+
+## Programmatic Usage
+
+```typescript
+import { parseSkillFile, search, trawl, createSkillliMcpServer } from 'skillli';
+
+// Parse a SKILL.md file
+const skill = await parseSkillFile('./SKILL.md');
+
+// Search the index
+const results = search(index, { query: 'code review', tags: ['security'] });
+
+// Trawl across sources
+const found = await trawl('kubernetes deployment', {
+  sources: ['registry', 'github', 'npm'],
+});
+
+// Create an MCP server
+const server = createSkillliMcpServer();
+```
+
+## License
+
+MIT