enki-runtime 0.1.4

A Rust-based agent mesh framework for building local and distributed AI agent systems
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102

# Web Browsing Agent Configuration

# Uses MCP (Model Context Protocol) for web browser automation

#

# Prerequisites:

# 1. Node.js and npm installed

# 2. Build with mcp feature: cargo build --features mcp

# 3. First run: npx will auto-install the MCP packages



name = "web-browsing-mesh"

version = "1.0"



# Coordinator configuration:

# Option 1: Designate an existing agent as coordinator

# coordinator = "web_browser"

#

# Option 2: Auto-generate a coordinator agent

auto_coordinator = true

coordinator_model = "ollama::gemma3:latest"  # Optional, defaults to first agent's model



# Web browser agent with Puppeteer MCP tools

[[agents]]

name = "web_browser"

model = "ollama::gemma3:latest"

system_prompt = """You are a web browsing assistant with full browser automation capabilities.

You can:
- Navigate to any URL
- Take screenshots of web pages
- Click on elements and fill out forms
- Execute JavaScript in the browser
- Capture and analyze console logs

When browsing, always:
1. Navigate to the requested URL first
2. Take a screenshot to understand the page layout
3. Perform the requested actions
4. Report what you found or did"""

temperature = 0.7

max_tokens = 4096



# Puppeteer browser automation MCP server

[[agents.mcp_tools]]

type = "stdio"

name = "puppeteer"

command = "npx"

args = ["-y", "@modelcontextprotocol/server-puppeteer"]



# Research agent that can browse the web and fetch content

[[agents]]

name = "researcher"

model = "ollama::gemma3:latest"

system_prompt = """You are a research assistant with web browsing capabilities.

You can:
- Navigate to websites and search engines
- Read and extract content from web pages
- Follow links to gather information
- Take screenshots to document findings

When researching:
1. Start with a search query or direct URL
2. Navigate through relevant pages
3. Extract and summarize key information
4. Provide sources for your findings"""

temperature = 0.5

max_tokens = 4096



# Puppeteer for full browser control

[[agents.mcp_tools]]

type = "stdio"

name = "browser"

command = "npx"

args = ["-y", "@modelcontextprotocol/server-puppeteer"]



# Fetch for lightweight HTTP requests

[[agents.mcp_tools]]

type = "stdio"

name = "fetch"

command = "npx"

args = ["-y", "@modelcontextprotocol/server-fetch"]



# Form automation agent

[[agents]]

name = "form_filler"

model = "ollama::gemma3:latest"

system_prompt = """You are a form automation assistant.

You can:
- Navigate to web forms
- Fill in form fields with provided data
- Select dropdowns and checkboxes
- Submit forms

Always take a screenshot before and after filling forms to verify your work."""

temperature = 0.3

max_tokens = 2048



[[agents.mcp_tools]]

type = "stdio"

name = "puppeteer"

command = "npx"

args = ["-y", "@modelcontextprotocol/server-puppeteer"]