@agent-workspace/mcp-server 0.2.0 → 0.4.0

package/README.md

@@ -0,0 +1,103 @@
+# @agent-workspace/mcp-server
+
+MCP server for the [Agent Workspace Protocol](https://github.com/marcoloco23/awp) — plug any AWP workspace into Claude Code, Cursor, or any MCP-compatible client.
+
+## Setup
+
+### Claude Code
+
+```bash
+claude mcp add awp-workspace -- npx @agent-workspace/mcp-server
+```
+
+### With a specific workspace path
+
+```bash
+AWP_WORKSPACE=/path/to/workspace npx @agent-workspace/mcp-server
+```
+
+### Generic MCP client
+
+The server communicates over stdio using the [Model Context Protocol](https://modelcontextprotocol.io). Point any MCP client at:
+
+```bash
+npx @agent-workspace/mcp-server
+```
+
+It will auto-discover the AWP workspace from the current directory (or use `AWP_WORKSPACE` env var).
+
+## Tools
+
+### Core Identity & Context
+
+| Tool | Description |
+|------|-------------|
+| `awp_read_identity` | Read agent identity (name, capabilities, DID) |
+| `awp_read_soul` | Read behavioral constraints (values, boundaries, governance) |
+| `awp_read_user` | Read human profile |
+| `awp_read_agents` | Read operations config (AGENTS.md) |
+| `awp_read_tools` | Read tools config (TOOLS.md) |
+| `awp_read_heartbeat` | Read heartbeat config (HEARTBEAT.md) |
+
+### Memory Operations
+
+| Tool | Description |
+|------|-------------|
+| `awp_read_memory` | Read memory (daily, longterm, or recent logs) |
+| `awp_write_memory` | Log structured memory entries to daily log |
+
+### Knowledge Artifacts (SMP)
+
+| Tool | Description |
+|------|-------------|
+| `awp_artifact_read` | Read a knowledge artifact by slug |
+| `awp_artifact_write` | Create or update a knowledge artifact with versioning |
+| `awp_artifact_list` | List all artifacts (optional tag filter) |
+| `awp_artifact_search` | Search artifacts by content, title, or tags |
+
+### Reputation (RDP)
+
+| Tool | Description |
+|------|-------------|
+| `awp_reputation_query` | Query reputation profiles with time-based decay applied |
+| `awp_reputation_signal` | Log reputation signals (uses EWMA with decay) |
+
+### Delegation Contracts (RDP)
+
+| Tool | Description |
+|------|-------------|
+| `awp_contract_create` | Create delegation contracts between agents |
+| `awp_contract_evaluate` | Evaluate contracts and generate reputation signals |
+| `awp_contract_list` | List all contracts with optional status filter |
+
+### Workspace Management
+
+| Tool | Description |
+|------|-------------|
+| `awp_workspace_status` | Workspace health check (files present, stats) |
+
+## How It Works
+
+The MCP server reads and writes AWP workspace files directly — no database, no daemon. Your agent's identity, memory, knowledge artifacts, and reputation profiles are just Markdown files with YAML frontmatter, stored in a directory and compatible with Git.
+
+When an AI agent connects via MCP, it can:
+- **Read** its identity, soul, and user context at the start of each session
+- **Write** memory entries as it works
+- **Create and update** knowledge artifacts with full provenance tracking
+- **Search** its accumulated knowledge
+- **Query and update** reputation profiles for itself and other agents
+- **Create and evaluate** delegation contracts for multi-agent coordination
+
+## Part of AWP
+
+This package is part of the [Agent Workspace Protocol](https://github.com/marcoloco23/awp) monorepo:
+
+| Package | Description |
+|---------|-------------|
+| [@agent-workspace/core](https://www.npmjs.com/package/@agent-workspace/core) | Types, constants, JSON schemas |
+| [@agent-workspace/cli](https://www.npmjs.com/package/@agent-workspace/cli) | CLI tool |
+| **@agent-workspace/mcp-server** | MCP server |
+
+## License
+
+Apache-2.0

package/dist/index.d.ts

@@ -1,3 +1,19 @@
 #!/usr/bin/env node
-export {};
+/**
+ * Validate that a path is within the workspace root (prevents directory traversal).
+ * Returns the normalized absolute path if valid, or throws an error.
+ * @internal Available for future use in tool handlers
+ */
+export declare function _validatePath(root: string, targetPath: string): string;
+/**
+ * Validate and sanitize a slug.
+ * Slugs must be lowercase alphanumeric with hyphens, not starting with hyphen.
+ * @internal Available for future use in tool handlers
+ */
+export declare function _validateSlug(slug: string): string;
+/**
+ * Read a file with size limit check.
+ * @internal Available for future use in tool handlers
+ */
+export declare function _safeReadFile(path: string): Promise<string>;
 //# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":""}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":";AAkCA;;;;GAIG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,GAAG,MAAM,CAUtE;AAED;;;;GAIG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAYlD;AAED;;;GAGG;AACH,wBAAsB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAMjE"}