Tools & Plugins
Tessera provides built-in tools and supports custom extensions via Python plugins and MCP servers.
Built-in Tools
File Operations
read_file- Read file contentswrite_file- Write to filesdelete_file- Delete fileslist_directory- List directory contents
Git Operations
git_status- Check repository statusgit_commit- Create commitsgit_push- Push to remotecreate_pr- Create pull requests
Web Operations
web_search- Search the webfetch_url- Retrieve web pagesscrape_page- Extract content
Code Execution
execute_python- Run Python codeexecute_shell- Run shell commandsrun_tests- Execute test suite
Tool Access Control
Configure which tools agents can use:
tools:
global:
strategy: "risk-based"
max_risk_level: "high"
builtin:
filesystem:
enabled: true
approval_required: ["write_file", "delete_file"]
execution:
enabled: true
approval_required: true # All execution needs approval
Python Plugins
Add custom tools via Python files:
Plugin Location
Plugin Example
from tessera.plugins import tool
@tool(
name="send_email",
description="Send email via Gmail API",
risk_level="medium",
approval_required=True
)
def send_gmail(to: str, subject: str, body: str) -> str:
# Implementation
return "Email sent"
Plugin Configuration
tools:
plugins:
discovery:
- "~/.config/tessera/plugins/*.py"
definitions:
- name: "gmail-tool"
file: "~/.config/tessera/plugins/gmail.py"
enabled: true
risk_level: "medium"
MCP Integration
Connect to Model Context Protocol servers:
tools:
mcp:
- name: "filesystem-mcp"
enabled: true
type: "stdio"
command: "npx"
args: ["-y", "@modelcontextprotocol/server-filesystem"]
env:
ALLOWED_DIRECTORIES: "./,~/.config/tessera"
Risk Levels
Tools are classified by risk:
- safe - Read-only operations
- low - Local modifications in safe paths
- medium - File writes, installations
- high - Deletions, system commands
- critical - Git push, deployments
Tool Development
Custom tools can be created by extending the base Tool class. Documentation for plugin development is coming soon.