Documentation Index
Fetch the complete documentation index at: https://docs.praison.ai/llms.txt
Use this file to discover all available pages before exploring further.
MCP Registry Bridge CLI
The Registry Bridge is an internal adapter that connects praisonaiagents.tools to the MCP server. While it doesn’t have dedicated CLI commands, its effects are visible through other CLI commands.
When the bridge is enabled, bridged tools appear in tool listings with the configured namespace prefix (default: praisonai.agents.).
# List all tools (includes bridged tools if available)
praisonai mcp list-tools
Available MCP Tools (75 of 75):
• praisonai.workflow.run
Execute a PraisonAI workflow
• praisonai.agents.web.search
Search the web (bridged from praisonaiagents)
• praisonai.agents.memory.store
Store data in memory (bridged from praisonaiagents)
...
# Search for bridged tools by namespace
praisonai mcp tools search "praisonai.agents" --json
# Get info on a bridged tool
praisonai mcp tools info praisonai.agents.web.search
Tool: praisonai.agents.web.search
Description: PraisonAI Agents tool: web.search
Annotations:
• readOnlyHint: True
• destructiveHint: False
• idempotentHint: False
• openWorldHint: True
• category: web
Bridge Status via Doctor
The doctor command shows bridge status:
Output:
MCP Server Health Check
=======================
✓ Core server: OK
✓ Tool registry: 50 tools registered
✓ Resource registry: 5 resources registered
✓ Prompt registry: 3 prompts registered
Bridge Status:
• praisonaiagents available: Yes
• Bridged tools: 25
• Namespace prefix: praisonai.agents.
Programmatic Bridge Control
While there’s no direct CLI for bridge control, you can use Python:
# Check bridge availability
python3 -c "
from praisonai.mcp_server.adapters.tools_bridge import is_bridge_available
print('Bridge available:', is_bridge_available())
"
# List bridged tools
python3 -c "
from praisonai.mcp_server.adapters.tools_bridge import (
is_bridge_enabled,
list_bridged_tools,
)
if is_bridge_enabled():
for tool in list_bridged_tools():
print(tool)
else:
print('Bridge not enabled')
"
Server with Bridge
When starting the MCP server, bridged tools are automatically registered if praisonaiagents is installed:
# Start server (bridge auto-enabled if available)
praisonai mcp serve
# Output includes bridge status
# [INFO] Registered 50 built-in tools
# [INFO] Bridge enabled: 25 tools from praisonaiagents
# [INFO] MCP server started on stdio
- Namespace prefix: Default
praisonai.agents.
- Description format: “PraisonAI Agents tool: ”
- Inferred annotations: Based on tool name patterns
#!/bin/bash
# Categorize tools by origin
echo "=== Built-in Tools ==="
praisonai mcp tools search --json | \
jq -r '.tools[] | select(.name | startswith("praisonai.agents") | not) | .name'
echo ""
echo "=== Bridged Tools ==="
praisonai mcp tools search "praisonai.agents" --json | \
jq -r '.tools[].name'
Troubleshooting
Bridge Not Available
# Check if praisonaiagents is installed
python3 -c "import praisonaiagents; print('OK')" 2>/dev/null || echo "Not installed"
# Install if needed
pip install praisonaiagents
# Verify bridge status
python3 -c "
from praisonai.mcp_server.adapters.tools_bridge import (
is_bridge_available,
is_bridge_enabled,
get_bridged_tool_count,
)
print('Available:', is_bridge_available())
print('Enabled:', is_bridge_enabled())
print('Tool count:', get_bridged_tool_count())
"
# Test a bridged tool
praisonai mcp tools schema praisonai.agents.web.search
# If the underlying module fails to load, you'll see:
# Error: Tool praisonai.agents.web.search failed to load: ...
| Aspect | Impact |
|---|
| Server startup | Minimal (metadata only) |
| Tool listing | No impact (lazy loading) |
| First tool call | Module import time |
| Subsequent calls | No overhead |
See Also
