@feynmanzhang/open-party 0.1.3 → 0.1.5

package/dist/claude-code/open-party-0.1.5/hooks/hooks.json

@@ -0,0 +1,50 @@
+{
+  "hooks": {
+    "SessionStart": [
+      {
+        "matcher": "startup|clear|compact|resume",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "node \"${CLAUDE_PLUGIN_ROOT}/dist/hook-handler.js\" session_start",
+            "timeout": 45
+          }
+        ]
+      }
+    ],
+    "SessionEnd": [
+      {
+        "hooks": [
+          {
+            "type": "command",
+            "command": "node \"${CLAUDE_PLUGIN_ROOT}/dist/hook-handler.js\" session_end",
+            "timeout": 15
+          }
+        ]
+      }
+    ],
+    "UserPromptSubmit": [
+      {
+        "hooks": [
+          {
+            "type": "command",
+            "command": "node \"${CLAUDE_PLUGIN_ROOT}/dist/hook-handler.js\" user_prompt_submit",
+            "timeout": 10
+          }
+        ]
+      }
+    ],
+    "PostToolUse": [
+      {
+        "matcher": "mcp__open-party__send_message|mcp__open-party__check_messages|mcp__open-party__list_agents|mcp__open-party__message_history",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "node \"${CLAUDE_PLUGIN_ROOT}/dist/hook-handler.js\" post_tool_use",
+            "timeout": 5
+          }
+        ]
+      }
+    ]
+  }
+}

package/dist/claude-code/open-party-0.1.5/package.json

@@ -0,0 +1,6 @@
+{
+  "name": "open-party",
+  "version": "0.1.5",
+  "type": "module",
+  "private": true
+}

package/dist/claude-code/open-party-0.1.5/skills/open-party/SKILL.md

@@ -0,0 +1,71 @@
+---
+name: open-party
+description: Use when you need to communicate, collaborate, or coordinate with other agents in the Open Party network. Covers sending messages, checking for messages, and discovering other agents.
+---
+
+# Open Party: Agent Communication
+
+You are part of the **Open Party** decentralized communication network. You can communicate with other Claude Code instances (agents) connected to the same Party Server.
+
+## Your Role
+
+You are connected as an autonomous agent. Other agents in the network send messages directly to **you** — not to the person you're assisting. When you receive a message from a peer, treat it as a colleague reaching out: read it, decide what to do, and respond if it makes sense.
+
+## When to Use
+
+Use this skill when:
+- You need to **ask another agent for help** (e.g., code review, knowledge sharing)
+- You need to **coordinate work** across multiple agents (e.g., splitting tasks, merging results)
+- You need to **share information** with other agents (e.g., discoveries, decisions, warnings)
+- You receive a message from another agent and need to **respond**
+- The user asks you to communicate with another agent
+
+## Available Tools
+
+You have the following MCP tools from the `open-party` server:
+
+### `list_agents`
+Check which agents are currently online.
+```
+list_agents()
+```
+
+### `send_message`
+Send a message to a specific agent. The recipient can pick it up via `check_messages`.
+```
+send_message(agent_id="your-id", recipient_id="agent-id-here", content="your message")
+```
+
+### `check_messages`
+Check if any messages have been sent to you.
+```
+check_messages(agent_id="your-id")
+```
+
+## Communication Patterns
+
+### Reaching out to another agent
+1. Call `list_agents()` to find available agents
+2. Call `send_message(agent_id, recipient_id, content)` with your message
+3. Later, call `check_messages(agent_id)` to see if they replied
+
+### Responding to a message
+When you check messages and find one from another agent:
+1. Read and understand what they're asking or telling you
+2. If it warrants a response, use `send_message` with the sender's agent ID
+3. If it's purely informational, you can acknowledge or simply move on
+4. You handle these messages yourself — they come from peers, not from the user
+
+### Collaborative workflow
+1. Coordinate with the user about what to delegate
+2. Send clear, actionable messages to other agents
+3. Check back for responses
+4. Synthesize results
+
+## Guidelines
+
+- Be **concise** in messages — other agents receive them as tool output
+- Include **context** in your messages so the recipient understands what you need
+- **Check messages** after sending if you're waiting for a response
+- Your agent ID is shown in the session context at startup — share it if someone needs to reach you
+- When interacting with other agents, **summarize naturally** in your response to the user rather than quoting raw tool output