kylebrodeur · Copilot · Feb 3, 2026 · Feb 3, 2026 · Feb 3, 2026 · Feb 3, 2026
diff --git a/.github/RELEASE_NOTES_v0.1.0.md b/.github/RELEASE_NOTES_v0.1.0.md
@@ -10,7 +10,7 @@ UACS is a universal system for discovering, translating, and managing AI agent c
 
 - **Format Translation**: Convert between `.cursorrules`, `.clinerules`, `.agents.md`, and other agent skill formats
 - **Context Compression**: Reduce token usage by 70%+ while preserving semantic meaning
-- **Marketplace Integration**: Search and install agent skills from multiple registries
+- **Package Management**: Search and install agent skills from multiple registries
 - **MCP Server**: Full Model Context Protocol implementation for integration with Claude Desktop, Cursor, and other MCP clients
 
 ## 📦 Installation Options
@@ -61,7 +61,7 @@ services:
 - Compress context for efficient token usage
 - Built-in validation and error checking
 
-### 🛍️ Marketplace
+### 🛍️ Package Management
 - Search across multiple skill registries
 - Install skills directly from CLI
 - Package validation and security scanning
@@ -90,8 +90,8 @@ uacs skills discover
 # Start MCP server
 uacs serve --port 8080
 
-# Search marketplace for agent skills
-uacs marketplace search "python testing"
+# Search packages for agent skills
+uacs packages search "python testing"
 ```
 
 ### Python API
@@ -148,7 +148,7 @@ UACS builds on:
 Upcoming features (Phase 3-6):
 - Windows and Linux binaries
 - Integration guides for Claude Desktop, Cursor, and Windsurf
-- Additional marketplace registries (Smithery.ai, GitHub Topics)
+- Additional package registries (Smithery.ai, GitHub Topics)
 - Enhanced memory system with semantic search
 - Performance optimizations and caching
 - Security hardening and audit

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -340,7 +340,7 @@ First public release of UACS - Universal Agent Context System for AI agent conte
 - **Context Management**: Discover, translate, and manage AI agent context across multiple formats
 - **Format Adapters**: Built-in support for `.cursorrules`, `.clinerules`, `.agents.md`, and agent skill formats
 - **Context Compression**: Intelligent token reduction (70%+ compression ratio) while preserving semantic meaning
-- **Marketplace Integration**: Search and install agent skills from multiple registries
+- **Package Management**: Search and install agent skills from multiple registries
 - **Memory System**: Simple in-memory storage for agent context and history
 
 #### MCP Server
@@ -442,7 +442,7 @@ This project builds on concepts from:
 - PyPI package distribution
 - Claude Desktop integration guide
 - Cursor and Windsurf integration guides
-- Additional marketplace registries
+- Additional package registries
 - Enhanced memory system with semantic search
 
 ---

diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -197,7 +197,7 @@ Use conventional commit style:
 feat: Add support for .windsurf format
 fix: Correct token counting in compression
 docs: Update ARCHITECTURE.md with new diagrams
-test: Add tests for marketplace caching
+test: Add tests for package caching
 refactor: Extract adapter registry to separate module
 ```
 
@@ -259,11 +259,11 @@ universal-agent-context/
 │   │   ├── unified_context.py   # UnifiedContextAdapter
 │   │   ├── shared_context.py    # SharedContextManager
 │   │   └── agent_context.py     # AgentContextBuilder
-│   ├── marketplace/             # Package discovery
-│   │   ├── marketplace.py       # Unified marketplace
-│   │   ├── repositories.py      # Git repo management
-│   │   ├── packages.py          # Package metadata
-│   │   └── cache.py             # Caching layer
+│   ├── packages/                 # Package discovery
+│   │   ├── manager.py            # Package manager
+│   │   ├── repositories.py       # Git repo management
+│   │   ├── packages.py           # Package metadata
+│   │   └── cache.py              # Caching layer
 │   ├── memory/                  # Persistent storage
 │   │   ├── simple_memory.py     # Key-value store
 │   │   └── retrieval.py         # Semantic search
@@ -272,7 +272,7 @@ universal-agent-context/
 │   ├── cli/                     # CLI commands
 │   │   ├── main.py              # Main app
 │   │   ├── skills.py            # Skills commands
-│   │   ├── marketplace_cli.py   # Marketplace commands
+│   │   ├── packages_cli.py      # Package commands
 │   │   └── ...
 │   └── utils/                   # Shared utilities
 │       ├── compression.py       # Compression helpers
@@ -281,12 +281,12 @@ universal-agent-context/
 ├── tests/                       # Test suite (mirrors src/)
 │   ├── adapters/
 │   ├── context/
-│   ├── marketplace/
+│   ├── packages/
 │   └── ...
 │
 ├── examples/                    # Usage examples
 │   ├── basic_context.py
-│   ├── marketplace_search.py
+│   ├── package_search.py
 │   └── ...
 │
 ├── docs/                        # Documentation
@@ -336,7 +336,7 @@ Include:
 - [Architecture Overview](docs/ARCHITECTURE.md)
 - [Library Guide](docs/LIBRARY_GUIDE.md)
 - [CLI Reference](docs/CLI_REFERENCE.md)
-- [Marketplace Guide](docs/MARKETPLACE.md)
+- [Package Management Guide](docs/features/PACKAGES.md)
 
 ---
 

diff --git a/docs/ARCHITECTURE.md b/docs/ARCHITECTURE.md
@@ -34,7 +34,7 @@ UACS is middleware for AI agent context management. It sits between your agent a
 │  Python API    │     CLI Tool         │      MCP Server               │
 │                │                      │                               │
 │  from uacs     │  $ uacs skills list  │  Server: uacs serve           │
-│  import UACS   │  $ uacs marketplace  │  Client: Claude Desktop       │
+│  import UACS   │  $ uacs packages     │  Client: Claude Desktop       │
 │                │    search "testing"  │          Cursor, Cline        │
 └────────┬───────┴──────────┬───────────┴──────────┬────────────────────┘
          │                  │                      │
@@ -234,8 +234,8 @@ results = store.search_memories(
 - Server: `uacs serve` (stdio-based MCP server)
 - Client: Claude Desktop connects via MCP protocol
 - Tools exposed:
-  - `search_marketplace` - Find skills/MCP servers
-  - `install_package` - Install from marketplace
+  - `search_packages` - Find skills/MCP servers
+  - `install_package` - Install from package registry
   - `get_context` - Build compressed context
   - `list_skills` - Show installed skills
   - `convert_format` - Translate between formats
@@ -545,11 +545,11 @@ All three use the same core (`uacs/api.py`), ensuring consistency.
 [dependencies]
 python = "^3.11"            # Modern Python with type hints
 pydantic = "^2.0"           # Data validation
-httpx = "^0.27"             # Async HTTP client (marketplace)
+httpx = "^0.27"             # Async HTTP client (packages)
 anyio = "^4.0"              # Async abstraction
 tiktoken = "^0.7"           # Token counting (OpenAI)
 zstandard = "^0.23"         # Compression (shared context)
-gitpython = "^3.1"          # Git operations (marketplace)
+gitpython = "^3.1"          # Git operations (packages)
 mcp = "^1.0"                # Model Context Protocol (server)
 typer = "^0.9"              # CLI framework
 rich = "^13.0"              # Terminal UI
@@ -566,7 +566,7 @@ bandit = "^1.7"             # Security scanning
 ### Why These Choices?
 
 **Pydantic:** Type-safe data validation, auto-documentation
-**httpx:** Modern async HTTP, needed for marketplace
+**httpx:** Modern async HTTP, needed for package management
 **tiktoken:** Accurate token counting for OpenAI-compatible models
 **zstandard:** Fast compression for context storage
 **typer + rich:** Beautiful CLI with minimal code

diff --git a/docs/CLI_REFERENCE.md b/docs/CLI_REFERENCE.md
@@ -1,6 +1,6 @@
 # UACS CLI Reference
 
-The `uacs` command-line interface provides tools for managing context, skills, marketplace packages, and memory.
+The `uacs` command-line interface provides tools for managing context, skills, packages, and memory.
 
 ## Global Options
 

diff --git a/docs/LIBRARY_GUIDE.md b/docs/LIBRARY_GUIDE.md
@@ -65,7 +65,7 @@ ctx.add_entry(
 history = ctx.get_compressed_context(max_tokens=2000)
 ```
 
-### Marketplace
+### Package Management
 
 Search and install skills programmatically.
 

diff --git a/docs/features/ADAPTERS.md b/docs/features/ADAPTERS.md
@@ -21,7 +21,7 @@ The AgentSkillAdapter validates all SKILL.md files against the Agent Skills spec
 - ✅ Directory name must match skill name
 - ✅ Optional fields: `compatibility` (max 500 chars), `license`, `metadata`, `allowed-tools`
 
-Invalid skills will generate warnings in `skills list` and errors during `marketplace install`.
+Invalid skills will generate warnings in `skills list` and errors during `packages install`.
 
 **Precedence rules:**
 When the same skill name exists in both directories, `.agent/` wins:
@@ -102,6 +102,6 @@ Converts the parsed content into a dictionary compatible with Google ADK's agent
 
 ## Related Documentation
 
-- [UACS Marketplace](./MARKETPLACE.md) - Installing and validating skills
+- [UACS Package Management](./PACKAGES.md) - Installing and validating skills
 - [Context Management](./CONTEXT.md) - Using parsed content
 - [Agent Skills Specification](https://agentskills.io/specification) - Official format
diff --git a/docs/features/INTEGRATIONS.md b/docs/features/INTEGRATIONS.md
@@ -14,7 +14,7 @@ UACS can be deployed as an MCP server in three ways:
 | **Python Package** | Development, customization | Python 3.11+ |
 | **Docker Container** | Team servers, isolation | Docker |
 
-All three methods expose the same 20+ MCP tools for skills management, context optimization, and marketplace integration.
+All three methods expose the same 20+ MCP tools for skills management, context optimization, and package integration.
 
 ---
 
@@ -191,10 +191,10 @@ All UACS integrations expose these MCP tools:
 - `unified_capabilities` - Get all unified capabilities
 - `unified_token_stats` - Get token usage across all sources
 
-### Marketplace Integration (3 tools)
-- `marketplace_search` - Search skills marketplace
-- `marketplace_install` - Install skill from marketplace
-- `marketplace_list_installed` - List installed marketplace skills
+### Package Management (3 tools)
+- `packages_search` - Search skills packages
+- `packages_install` - Install skill from package registry
+- `packages_list_installed` - List installed packages
 
 ### Project Validation (1 tool)
 - `project_validate` - Validate AGENTS.md and skills configuration
@@ -238,10 +238,10 @@ Customize UACS behavior across all clients:
 
 **All Clients**:
 ```
-Search the marketplace for "testing" skills and show me the results.
+Search the packages for "testing" skills and show me the results.
 ```
 
-UACS will query multiple marketplace sources and return ranked results.
+UACS will query multiple package sources and return ranked results.
 
 ### 2. Context Optimization
 
@@ -288,7 +288,7 @@ Ensures your project follows best practices.
 |---------------|---------|-----|
 | Skills list/show | 10-30ms | 100ms |
 | Context stats | 20-50ms | 150ms |
-| Marketplace search | 50-150ms | 500ms |
+| Package search | 50-150ms | 500ms |
 | Context compression | 100-300ms | 1000ms |
 
 ### Memory Usage
@@ -455,7 +455,7 @@ Point UACS to your organization's private skills:
 {
   "env": {
     "UACS_SKILLS_REPO": "https://github.com/yourorg/skills.git",
-    "UACS_MARKETPLACE_CACHE": "86400"
+    "UACS_PACKAGES_CACHE": "86400"
   }
 }
 ```
@@ -524,7 +524,7 @@ UACS implements **MCP v1.0** specification:
 
 When running UACS locally (binary/Python):
 
-- ✅ No network traffic (except marketplace features)
+- ✅ No network traffic (except package features)
 - ✅ All data stays on your machine
 - ✅ State directory is user-owned
 - ⚠️ Binary must be trusted (verify checksums)
@@ -572,7 +572,7 @@ If you're currently using UACS CLI and want to integrate with an MCP client:
    ```bash
    # CLI still works alongside MCP
    uacs skills list
-   uacs marketplace search "testing"
+   uacs packages search "testing"
    ```
 
 ### From Other Context Systems
@@ -658,7 +658,7 @@ A: Yes, all 20+ tools work in all tested clients (Claude Desktop, Cursor, Windsu
 A: Standalone binary (stdio) has the lowest latency (~100ms startup, 10-50ms per tool call).
 
 **Q: Can I run UACS without internet?**  
-A: Yes, all core features work offline. Only marketplace search/install requires internet.
+A: Yes, all core features work offline. Only package search/install requires internet.
 
 **Q: How do I update UACS?**  
 A: Reinstall using your chosen method. Your state directory is preserved automatically.
@@ -667,7 +667,7 @@ A: Reinstall using your chosen method. Your state directory is preserved automat
 A: Yes, UACS is open source (MIT license) and free to use.
 
 **Q: Does UACS send data to external servers?**  
-A: No, except for marketplace features which query public repositories (optional).
+A: No, except for package features which query public repositories (optional).
 
 **Q: Can I contribute a new integration?**  
 A: Absolutely! Test UACS with your favorite MCP client and submit a PR with your integration guide.
@@ -678,7 +678,7 @@ A: Absolutely! Test UACS with your favorite MCP client and submit a PR with your
 
 Now that you've chosen your integration, explore:
 
-1. **[Skills Marketplace](../MARKETPLACE.md)** - Discover 100+ pre-built skills
+1. **[Package Management](PACKAGES.md)** - Discover 100+ pre-built skills
 2. **[Context Management](CONTEXT.md)** - Deep dive into context optimization  
 3. **[AGENTS.md Format](ADAPTERS.md)** - Multi-agent coordination
 4. **[Library Guide](../LIBRARY_GUIDE.md)** - Use UACS programmatically in Python

diff --git a/docs/integrations/CLAUDE_DESKTOP.md b/docs/integrations/CLAUDE_DESKTOP.md
@@ -1,6 +1,6 @@
 # UACS Integration with Claude Desktop
 
-**Universal Agent Context System (UACS)** can be integrated with Claude Desktop as an MCP (Model Context Protocol) server, providing powerful context management, skills marketplace, and multi-format translation capabilities directly in your Claude conversations.
+**Universal Agent Context System (UACS)** can be integrated with Claude Desktop as an MCP (Model Context Protocol) server, providing powerful context management, package management, and multi-format translation capabilities directly in your Claude conversations.
 
 ## Quick Start
 
@@ -301,7 +301,7 @@ What MCP tools do you have available?
 
 Claude should list UACS tools including:
 - `skills_list` - List all available agent skills
-- `marketplace_search` - Search skills marketplace
+- `packages_search` - Search skills packages
 - `context_stats` - Get context statistics
 - `unified_build_prompt` - Build complete agent prompt
 - And 20+ more tools...
@@ -315,9 +315,9 @@ Try these example prompts in Claude:
 Use the skills_list tool to show me what skills are available.
 ```
 
-#### Search the Marketplace
+#### Search Packages
 ```
-Search the skills marketplace for "testing" related skills.
+Search the package registry for "testing" related skills.
 ```
 
 #### Get Context Statistics
@@ -334,9 +334,9 @@ Use skills_test_trigger to see which skill would handle the query "review my cod
 
 Here's a complete example of using UACS with Claude Desktop:
 
-**You**: "Search the marketplace for Python testing tools"
+**You**: "Search the package registry for Python testing tools"
 
-**Claude**: Uses `marketplace_search` tool with query="python testing"
+**Claude**: Uses `packages_search` tool with query="python testing"
 
 **Result**:
 ```json
@@ -388,10 +388,10 @@ UACS exposes 20+ MCP tools organized by category:
 - `unified_capabilities` - Get all unified capabilities
 - `unified_token_stats` - Get token usage across all sources
 
-### Marketplace Integration
-- `marketplace_search` - Search skills marketplace
-- `marketplace_install` - Install skill from marketplace
-- `marketplace_list_installed` - List installed marketplace skills
+### Package Management
+- `packages_search` - Search skills packages
+- `packages_install` - Install skill from package registry
+- `packages_list_installed` - List installed packages
 
 ### Project Validation
 - `project_validate` - Validate AGENTS.md and skills configuration
@@ -607,7 +607,7 @@ journalctl --user -u claude-desktop -f
 
 ### Token Processing
 - **Context compression**: 10,000 tokens/second
-- **Marketplace search**: <100ms typical
+- **Package search**: <100ms typical
 - **Skill validation**: <50ms per skill
 
 ---
@@ -691,7 +691,7 @@ Run different UACS instances for different projects:
 A: Yes! UACS implements the standard MCP protocol and works with any MCP-compatible client (Claude Desktop, Cursor, Windsurf, etc.).
 
 **Q: Does UACS require internet access?**  
-A: No, UACS runs locally. Internet is only needed for marketplace features (optional).
+A: No, UACS runs locally. Internet is only needed for package features (optional).
 
 **Q: Can I customize which tools are exposed?**  
 A: Not yet, but this feature is planned. Track [issue #XX](https://github.com/kylebrodeur/universal-agent-context/issues).
@@ -705,7 +705,7 @@ A: Reinstall using your chosen method (binary/Docker/Python). Your state directo
 
 Now that you have UACS integrated with Claude Desktop, explore:
 
-1. **[Skills Marketplace](../MARKETPLACE.md)** - Discover and install pre-built skills
+1. **[Package Management](../features/PACKAGES.md)** - Discover and install pre-built skills
 2. **[Context Management](../CONTEXT.md)** - Learn about advanced context features
 3. **[AGENTS.md](../../docs/ADAPTERS.md)** - Multi-agent coordination
 4. **[Library Guide](../LIBRARY_GUIDE.md)** - Use UACS in Python code